Research/specclaw - #195
Closed
crispyberry wants to merge 2 commits into
Closed
Conversation
…on plan (EN/ZH) Research documents for the MCU-hosted agent loop work: - mimiclaw-system-report(.md/-zh.md): system report with evaluation plan - specclaw-research-proposal(.md/-zh.md): guarded speculative execution proposal - proposal-speculative-execution.md: engineering implementation plan Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
📝 WalkthroughWalkthroughThe pull request adds MimiClaw system reports, SpecClaw guarded-speculation proposals, and a LaTeX thesis scaffold with chapter structures, document wiring, appendices, and bibliography entries. ChangesMimiClaw system reports
SpecClaw research proposals
Thesis document scaffold
Estimated code review effort: 3 (Moderate) | ~20 minutes 🚥 Pre-merge checks | ✅ 4 | ❌ 1❌ Failed checks (1 inconclusive)
✅ Passed checks (4 passed)
✨ Finishing Touches🧪 Generate unit tests (beta)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 19
🧹 Nitpick comments (2)
docs/specclaw-research-proposal.md (1)
97-103: 🎯 Functional Correctness | 🔵 Trivial | 🏗️ Heavy liftSpecify how “matched task success” is enforced.
B2 and unsafe guard configurations may have materially different success rates. State whether configurations are tuned to a common success target, compared on a success–cost frontier, or evaluated with success as a constraint; otherwise the headline comparison is not reproducible.
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the rest with a brief reason, keep changes minimal, and validate. In `@docs/specclaw-research-proposal.md` around lines 97 - 103, Clarify in RQ1 how “matched task success” is enforced across B2 and unsafe guard configurations. Specify whether configurations are tuned to a common success target, compared on a success–cost frontier, or evaluated with success as a constraint, and use that policy consistently for the headline comparison.docs/specclaw-research-proposal-zh.md (1)
99-105: 🎯 Functional Correctness | 🔵 Trivial | 🏗️ Heavy lift明确“成功率持平”的实验匹配方法。
不同 baseline 可能具有不同任务成功率。请说明是调参到共同成功率、比较成功率—成本前沿，还是将成功率作为约束；否则该比较难以复现。
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the rest with a brief reason, keep changes minimal, and validate. In `@docs/specclaw-research-proposal-zh.md` around lines 99 - 105, 在 RQ1 的实验设计中明确“成功率持平”的匹配方法：规定如何处理不同 baseline 的成功率，例如调参至共同成功率、比较成功率—成本前沿，或将成功率设为约束。说明具体采用的方案及匹配标准，确保云调用、能耗和延迟比较可复现。
🤖 Prompt for all review comments with AI agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Inline comments:
In `@docs/mimiclaw-system-report.md`:
- Line 80: The RAM ledger is inconsistent with the reconciled task count. Update
the RAM summary in docs/mimiclaw-system-report.md (line 80) and its
corresponding Chinese passage in docs/mimiclaw-system-report-zh.md (line 79) to
use the corrected total and revise the “under 60 KB” claim consistently in both
documents.
- Line 11: The OS terminology is inaccurate in both reports:
docs/mimiclaw-system-report.md:11-11 and docs/mimiclaw-system-report-zh.md:10-10
must describe MimiClaw as a FreeRTOS-based embedded runtime without a
general-purpose OS or process model. Replace the English “bare-metal” and “no
operating system” wording and the Chinese “裸机” and “没有操作系统” wording accordingly,
while preserving the surrounding claims.
- Line 172: Clarify the per-call overhead wording in the report’s ReAct cost
description: in docs/mimiclaw-system-report.md:172-172, replace “fixed cost
independent of tokens” with wording that specifies fixed latency and energy
overhead; apply the equivalent wording correction in
docs/mimiclaw-system-report-zh.md:171-171.
In `@docs/proposal-speculative-execution.md`:
- Around line 59-81: Define the canonical timeout field and its scope in the
structured script contract, ensuring it matches the parser/evaluator behavior.
Update the corresponding schema examples and descriptions in
docs/proposal-speculative-execution.md (59-81),
docs/specclaw-research-proposal.md (64-83), and
docs/specclaw-research-proposal-zh.md (66-85) so all three remain semantically
identical.
- Around line 165-167: Update the “6.5 Ablations” section’s D_irrev list to
explicitly label D_irrev = ∞ as an unsafe, safety-off control. Clarify that this
condition disables the irreversible-action safeguard and must be excluded from
any safety-guarantee claims, while retaining it only for the ablation
comparison.
- Line 85: Update every fenced code block in
docs/proposal-speculative-execution.md, including the pseudocode, state
snapshot, and directory tree sections, to specify an appropriate language such
as text after the opening fence. Apply the same change to the additional fences
at the referenced locations while preserving their contents.
- Around line 85-98: Qualify the protocol’s “never mis-executed” guarantee and
define recovery behavior when a post-guard fails after an action has executed:
specify safe-stop, compensation, or operator intervention before replanning.
Apply the same recovery semantics and guarantee boundaries in
docs/proposal-speculative-execution.md lines 85-98,
docs/specclaw-research-proposal.md lines 87-93, and
docs/specclaw-research-proposal-zh.md lines 89-95, preserving the existing
pre-guard, D_irrev, and gpio_policy requirements.
- Around line 123-128: Update the RQ1 hypothesis in
docs/proposal-speculative-execution.md at lines 123-128 to use the ideal ceil(n
/ L) call baseline instead of 1/L, while reporting measured calls and replans
separately. Apply the same correction in docs/specclaw-research-proposal.md at
lines 20-22 and docs/specclaw-research-proposal-zh.md at lines 22-24, explicitly
noting that failed scripts add calls beyond the ideal ceil(n / L).
- Around line 64-74: Update the protocol examples and guidance so
irreversibility is determined on-device from trusted tool metadata, with unknown
classifications rejected rather than accepted from the cloud-provided
irreversible field. Apply this rule in docs/proposal-speculative-execution.md
(lines 64-74), docs/specclaw-research-proposal.md (lines 70-75) to prevent
malformed values bypassing D_irrev, and docs/specclaw-research-proposal-zh.md
(lines 72-77) using the equivalent device-side authority wording.
In `@docs/specclaw-research-proposal-zh.md`:
- Around line 123-125: 在“消融”段落的 D_irrev 配置说明中，明确将 D_irrev = ∞
标记为关闭安全阀的对照组。说明该配置仅用于对照，不属于安全运行配置，且不得用于支持守卫安全性的结论；其余消融设置保持不变。
In `@docs/specclaw-research-proposal.md`:
- Around line 121-123: Update the Ablations section’s D_irrev comparison to
explicitly label D_irrev = ∞ as a safety-off baseline for unbounded irreversible
speculation. Separate it from the protocol’s safe operating envelope and exclude
it from guarded-safety claims.
In `@docs/thesis/chapters/ch1-introduction.tex`:
- Around line 24-27: Update the RQ-to-chapter mapping statement in the
introduction comments to allow each research question to map to one or more
results chapters, consistent with RQ1’s mapping to Chapters 3 and 5.
In `@docs/thesis/chapters/ch2-background.tex`:
- Around line 2-3: Update the Chapter 2 header’s stated literature count to
match the seven literature sections, including “Evaluating Agent Systems,” while
preserving the existing chapter structure.
In `@docs/thesis/main.tex`:
- Around line 70-72: Update the word-count declaration near the contents-page
setup to use the actual final word count instead of “XX,XXX”, and place it at
the bottom of the contents page. Remove the \cleardoublepage and standalone-page
behavior while preserving the existing formatting.
- Around line 103-128: Update the four copyright notes in the itemize block to
exactly match the mandated March 2026 University of Manchester policy wording,
including the required pronoun wording and capitalization/article in the final
note. Preserve the existing four-item structure and do not paraphrase the
official text.
- Around line 75-76: Insert \phantomsection immediately before each
\addcontentsline command for the unnumbered Abstract, Declaration, Copyright
Statement, and Acknowledgements entries, preserving their existing chapter and
ToC entries.
- Around line 153-154: Update the bibliography setup using \bibliographystyle
and \bibliography so the unnumbered bibliography is added to the table of
contents. Use the project’s existing TOC-aware package or add an equivalent
hyperlink-safe \addcontentsline wrapper, preserving the current natbib
bibliography behavior.
- Around line 46-61: Adjust the title-page flow after \end{titlepage} so the
following page begins at page 2, preserving continuous thesis pagination with
the title page counted as page 1. Add the page-counter update immediately after
the titlepage environment.
In `@docs/thesis/references.bib`:
- Line 13: Normalize the author metadata in docs/thesis/references.bib at lines
13-13, 43-44, and 180-180: use Xueyu Hu for the AAAI proceedings entry, replace
Anonymous with Jabbarvaziri, Faramarz, and replace the remaining placeholder
with Hu, Yuanzhe, Wang, Yu, and McAuley, Julian.
---
Nitpick comments:
In `@docs/specclaw-research-proposal-zh.md`:
- Around line 99-105: 在 RQ1 的实验设计中明确“成功率持平”的匹配方法：规定如何处理不同 baseline
的成功率，例如调参至共同成功率、比较成功率—成本前沿，或将成功率设为约束。说明具体采用的方案及匹配标准，确保云调用、能耗和延迟比较可复现。
In `@docs/specclaw-research-proposal.md`:
- Around line 97-103: Clarify in RQ1 how “matched task success” is enforced
across B2 and unsafe guard configurations. Specify whether configurations are
tuned to a common success target, compared on a success–cost frontier, or
evaluated with success as a constraint, and use that policy consistently for the
headline comparison.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: defaults
Review profile: CHILL
Plan: Pro
Run ID: a7a7d050-c48e-4c59-b16e-656c3d7568fa
📒 Files selected for processing (17)
docs/mimiclaw-system-report-zh.mddocs/mimiclaw-system-report.mddocs/proposal-speculative-execution.mddocs/specclaw-research-proposal-zh.mddocs/specclaw-research-proposal.mddocs/thesis/chapters/appendix.texdocs/thesis/chapters/ch1-introduction.texdocs/thesis/chapters/ch2-background.texdocs/thesis/chapters/ch3-system.texdocs/thesis/chapters/ch4-methodology.texdocs/thesis/chapters/ch5-characterization.texdocs/thesis/chapters/ch6-context-compaction.texdocs/thesis/chapters/ch7-skills.texdocs/thesis/chapters/ch8-speculative-execution.texdocs/thesis/chapters/ch9-conclusion.texdocs/thesis/main.texdocs/thesis/references.bib
|
|
||
| ## Abstract | ||
|
|
||
| LLM agent runtimes have been shrinking steadily: from OpenClaw's 400,000+ lines of TypeScript, to nanobot's 4,000 lines of Python, to sub-10 MB Go and Rust implementations such as PicoClaw and ZeroClaw. This trajectory stops, collectively, at one boundary: all of these systems require Linux — processes, a hierarchical filesystem, a dynamic language runtime, a package manager. MimiClaw crosses that boundary. It implements a complete agent runtime in roughly 8,300 lines of C on a bare-metal ESP32-S3 with no operating system, no processes, and 512 KB of on-chip SRAM: a ReAct tool loop, persistent memory managed by the model itself, an extensible skill system, multi-channel messaging, scheduled and heartbeat-driven autonomous behavior, and deterministic safety guards for hardware actuation, with reasoning delegated to a cloud LLM API. This report argues for the central claim this system demonstrates: **the essence of an agent runtime does not depend on an operating system.** We show how each Linux-era agent primitive is rebuilt on embedded primitives — processes become FreeRTOS tasks, async frameworks become message queues, plugin systems become Markdown data files, vector memory stores become plain-text files the model reads and writes itself — and we provide a complete resource ledger for this reconstruction, four engineering principles that recur throughout the design, and an honest account of the open problems it exposes. |
There was a problem hiding this comment.
🎯 Functional Correctness | 🟠 Major | ⚡ Quick win
Correct the OS terminology in both reports. The implementation is FreeRTOS-based, not OS-less; describe it as an embedded runtime without a general-purpose OS or process model.
docs/mimiclaw-system-report.md#L11-L11: revise “no operating system” and “bare-metal” wording.docs/mimiclaw-system-report-zh.md#L10-L10: revise “没有操作系统” and “裸机” wording.
📍 Affects 2 files
docs/mimiclaw-system-report.md#L11-L11(this comment)docs/mimiclaw-system-report-zh.md#L10-L10
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
In `@docs/mimiclaw-system-report.md` at line 11, The OS terminology is inaccurate
in both reports: docs/mimiclaw-system-report.md:11-11 and
docs/mimiclaw-system-report-zh.md:10-10 must describe MimiClaw as a
FreeRTOS-based embedded runtime without a general-purpose OS or process model.
Replace the English “bare-metal” and “no operating system” wording and the
Chinese “裸机” and “没有操作系统” wording accordingly, while preserving the surrounding
claims.
|
|
||
| **Flash (16 MB):** 2 × 2 MB OTA application partitions + 12 MB SPIFFS (memory/sessions/skills/config). The firmware itself is about 2 MB — a complete agent runtime in the code footprint of a mid-sized npm dependency. | ||
|
|
||
| **RAM:** static task stacks total about 56 KB of on-chip SRAM (agent 24 KB; three channel/dispatch tasks at 12 KB each; CLI 4 KB); the agent's 56 KB of working buffers (16 + 32 + 8 KB) reside entirely in PSRAM. Roughly stated: **the architectural part of an agent runs in under 60 KB of on-chip memory**, leaving the rest to the network stack and TLS. |
There was a problem hiding this comment.
🎯 Functional Correctness | 🟠 Major | ⚡ Quick win
Fix the inconsistent RAM totals.
docs/mimiclaw-system-report.md#L80-L80: update the 56 KB total and “under 60 KB” claim after reconciling the task count.docs/mimiclaw-system-report-zh.md#L79-L79: apply the same corrected ledger.
📍 Affects 2 files
docs/mimiclaw-system-report.md#L80-L80(this comment)docs/mimiclaw-system-report-zh.md#L79-L79
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
In `@docs/mimiclaw-system-report.md` at line 80, The RAM ledger is inconsistent
with the reconciled task count. Update the RAM summary in
docs/mimiclaw-system-report.md (line 80) and its corresponding Chinese passage
in docs/mimiclaw-system-report-zh.md (line 79) to use the corrected total and
revise the “under 60 KB” claim consistently in both documents.
|
|
||
| ## 8. Limitations and Next Steps | ||
|
|
||
| Honest boundaries matter as much as claims. **First**, the device currently depends on the cloud at every step: ReAct issues one API call per step, each paying a fixed cost independent of tokens (radio wake-up, TLS handshake, round trip), and during an outage the loop survives but reasoning halts. This is the direct motivation for our follow-up work on guarded speculative execution (see `specclaw-research-proposal.md`), for which the bus, tool allowlist, and file persistence described here are precisely the prerequisites. **Second**, this report provides a design and a static ledger; the measurement plan in Section 6 is specified but not yet executed, and carrying it out is the work required to move from a systems report to a systems paper. **Third**, session slimming and fixed budgets control context volume, but growth of the memory files themselves still relies on the model's discipline; there is no firmware-enforced compaction. |
There was a problem hiding this comment.
🎯 Functional Correctness | 🟡 Minor | ⚡ Quick win
Clarify that the fixed cost is per-call overhead, not token-independent billing.
docs/mimiclaw-system-report.md#L172-L172: replace “fixed cost independent of tokens” with fixed latency/energy overhead.docs/mimiclaw-system-report-zh.md#L171-L171: apply the equivalent wording correction.
📍 Affects 2 files
docs/mimiclaw-system-report.md#L172-L172(this comment)docs/mimiclaw-system-report-zh.md#L171-L171
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
In `@docs/mimiclaw-system-report.md` at line 172, Clarify the per-call overhead
wording in the report’s ReAct cost description: in
docs/mimiclaw-system-report.md:172-172, replace “fixed cost independent of
tokens” with wording that specifies fixed latency and energy overhead; apply the
equivalent wording correction in docs/mimiclaw-system-report-zh.md:171-171.
Comment on lines
+59
to
+81
| Each cloud call returns (enforced via structured output): | ||
|
|
||
| ```json | ||
| { | ||
| "state_update": "relay=on; last_temp=27.3; goal_phase=cooling", | ||
| "script": [ | ||
| { | ||
| "step": 1, | ||
| "action": {"tool": "gpio", "args": {"pin": 5, "level": 1}}, | ||
| "irreversible": false, | ||
| "guard": { | ||
| "pre": [{"var": "temp_c", "op": "gt", "val": 26.0}], | ||
| "post": [{"var": "temp_c", "op": "lt", "val": 26.5, "within_s": 120}, | ||
| {"var": "relay_current_ma", "op": "in", "lo": 80, "hi": 200, "within_s": 2}] | ||
| }, | ||
| "on_fail": "replan" | ||
| } | ||
| ], | ||
| "done_when": [{"var": "temp_c", "op": "lt", "val": 25.0}] | ||
| } | ||
| ``` | ||
|
|
||
| Guard language (deliberately minimal in v1): numeric comparisons `gt/lt/in`, discrete equality `eq`, rate-of-change `delta_lt`, time bounds `within_s`, AND composition only (no OR). The limited expressiveness is a feature, not a bug — decidable, evaluable in ~100 lines of C, statically auditable. |
There was a problem hiding this comment.
🗄️ Data Integrity & Integration | 🟠 Major | ⚡ Quick win
The documented script contract omits its promised timeout field.
docs/proposal-speculative-execution.md#L59-L81: define a canonical timeout field and its scope.docs/specclaw-research-proposal.md#L64-L83: align the structured-output example with the parser/evaluator contract.docs/specclaw-research-proposal-zh.md#L66-L85: keep the Chinese schema description semantically identical.
📍 Affects 3 files
docs/proposal-speculative-execution.md#L59-L81(this comment)docs/specclaw-research-proposal.md#L64-L83docs/specclaw-research-proposal-zh.md#L66-L85
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
In `@docs/proposal-speculative-execution.md` around lines 59 - 81, Define the
canonical timeout field and its scope in the structured script contract,
ensuring it matches the parser/evaluator behavior. Update the corresponding
schema examples and descriptions in docs/proposal-speculative-execution.md
(59-81), docs/specclaw-research-proposal.md (64-83), and
docs/specclaw-research-proposal-zh.md (66-85) so all three remain semantically
identical.
Comment on lines
+64
to
+74
| "script": [ | ||
| { | ||
| "step": 1, | ||
| "action": {"tool": "gpio", "args": {"pin": 5, "level": 1}}, | ||
| "irreversible": false, | ||
| "guard": { | ||
| "pre": [{"var": "temp_c", "op": "gt", "val": 26.0}], | ||
| "post": [{"var": "temp_c", "op": "lt", "val": 26.5, "within_s": 120}, | ||
| {"var": "relay_current_ma", "op": "in", "lo": 80, "hi": 200, "within_s": 2}] | ||
| }, | ||
| "on_fail": "replan" |
There was a problem hiding this comment.
🔒 Security & Privacy | 🟠 Major | 🏗️ Heavy lift
Irreversibility must be determined locally, not by the cloud script.
docs/proposal-speculative-execution.md#L64-L74: derive the safety class from trusted tool metadata or reject unknown classifications.docs/specclaw-research-proposal.md#L70-L75: prevent a malformedirreversiblevalue from bypassingD_irrev.docs/specclaw-research-proposal-zh.md#L72-L77: apply the same device-side authority rule in the Chinese protocol.
📍 Affects 3 files
docs/proposal-speculative-execution.md#L64-L74(this comment)docs/specclaw-research-proposal.md#L70-L75docs/specclaw-research-proposal-zh.md#L72-L77
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
In `@docs/proposal-speculative-execution.md` around lines 64 - 74, Update the
protocol examples and guidance so irreversibility is determined on-device from
trusted tool metadata, with unknown classifications rejected rather than
accepted from the cloud-provided irreversible field. Apply this rule in
docs/proposal-speculative-execution.md (lines 64-74),
docs/specclaw-research-proposal.md (lines 70-75) to prevent malformed values
bypassing D_irrev, and docs/specclaw-research-proposal-zh.md (lines 72-77) using
the equivalent device-side authority wording.
Comment on lines
+70
to
+72
| % Word count: policy requires it declared; put it on the contents page or here. | ||
| \cleardoublepage | ||
| \noindent Word count: \texttt{XX,XXX} |
There was a problem hiding this comment.
🗄️ Data Integrity & Integration | 🟠 Major | ⚡ Quick win
Place the final word count at the bottom of the contents page.
This currently creates a separate page and leaves XX,XXX as a placeholder. The current University of Manchester policy requires the final word count at the bottom of the contents page. (documents.manchester.ac.uk)
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
In `@docs/thesis/main.tex` around lines 70 - 72, Update the word-count declaration
near the contents-page setup to use the actual final word count instead of
“XX,XXX”, and place it at the bottom of the contents page. Remove the
\cleardoublepage and standalone-page behavior while preserving the existing
formatting.
Source: MCP tools
Comment on lines
+75
to
+76
| \chapter*{Abstract} | ||
| \addcontentsline{toc}{chapter}{Abstract} |
There was a problem hiding this comment.
🎯 Functional Correctness | 🟡 Minor | ⚡ Quick win
🧩 Analysis chain
🏁 Script executed:
#!/bin/bash
set -euo pipefail
# Inspect the thesis file around the cited lines
wc -l docs/thesis/main.tex
sed -n '1,180p' docs/thesis/main.tex
# Locate hyperref usage in the thesis docs
rg -n "\\\\usepackage\\{hyperref\\}|\\\\phantomsection|\\\\addcontentsline\\{toc\\}" docs/thesis -SRepository: memovai/mimiclaw
Length of output: 7293
Add \phantomsection before each unnumbered ToC entry. hyperref needs an explicit anchor here; otherwise the Abstract, Declaration, Copyright Statement, and Acknowledgements links can jump to the previous page/heading. Also apply this at lines 92-93, 99-100, and 132-133.
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
In `@docs/thesis/main.tex` around lines 75 - 76, Insert \phantomsection
immediately before each \addcontentsline command for the unnumbered Abstract,
Declaration, Copyright Statement, and Acknowledgements entries, preserving their
existing chapter and ToC entries.
Source: MCP tools
Comment on lines
+103
to
+128
| \begin{itemize} | ||
| \item[i.] The author of this thesis (including any appendices and/or | ||
| schedules to this thesis) owns certain copyright or related rights in it | ||
| (the ``Copyright'') and s/he has given The University of Manchester certain | ||
| rights to use such Copyright, including for administrative purposes. | ||
| \item[ii.] Copies of this thesis, either in full or in extracts and whether | ||
| in hard or electronic copy, may be made \emph{only} in accordance with the | ||
| Copyright, Designs and Patents Act 1988 (as amended) and regulations issued | ||
| under it or, where appropriate, in accordance with licensing agreements | ||
| which the University has from time to time. This page must form part of any | ||
| such copies made. | ||
| \item[iii.] The ownership of certain Copyright, patents, designs, trademarks | ||
| and other intellectual property (the ``Intellectual Property'') and any | ||
| reproductions of copyright works in the thesis, for example graphs and | ||
| tables (``Reproductions''), which may be described in this thesis, may not | ||
| be owned by the author and may be owned by third parties. Such Intellectual | ||
| Property and Reproductions cannot and must not be made available for use | ||
| without the prior written permission of the owner(s) of the relevant | ||
| Intellectual Property and/or Reproductions. | ||
| \item[iv.] Further information on the conditions under which disclosure, | ||
| publication and commercialisation of this thesis, the Copyright and any | ||
| Intellectual Property and/or Reproductions described in it may take place | ||
| is available in the University IP Policy, in any relevant Thesis | ||
| restriction declarations deposited in the University Library, the | ||
| University Library's regulations and in The University's policy on | ||
| Presentation of Theses. |
There was a problem hiding this comment.
🔒 Security & Privacy | 🟠 Major | ⚡ Quick win
Use the mandated copyright wording verbatim.
The current policy explicitly requires the four copyright notes “as written below.” The s/he wording and the final sentence’s capitalization/article differ from the March 2026 policy text; replace these passages with the official wording rather than paraphrasing it. (documents.manchester.ac.uk)
Representative corrections
- (``Copyright'') and s/he has given The University of Manchester certain
+ (``Copyright'') and they have given the University of Manchester certain
...
- is available in the University IP Policy, ... and in The University's policy on
- Presentation of Theses.
+ is available in the University IP Policy, ... and in the University's policy on
+ the Presentation of Theses.📝 Committable suggestion
‼️ IMPORTANT
Carefully review the code before committing. Ensure that it accurately replaces the highlighted code, contains no missing lines, and has no issues with indentation. Thoroughly test & benchmark the code to ensure it meets the requirements.
Suggested change
| \begin{itemize} | |
| \item[i.] The author of this thesis (including any appendices and/or | |
| schedules to this thesis) owns certain copyright or related rights in it | |
| (the ``Copyright'') and s/he has given The University of Manchester certain | |
| rights to use such Copyright, including for administrative purposes. | |
| \item[ii.] Copies of this thesis, either in full or in extracts and whether | |
| in hard or electronic copy, may be made \emph{only} in accordance with the | |
| Copyright, Designs and Patents Act 1988 (as amended) and regulations issued | |
| under it or, where appropriate, in accordance with licensing agreements | |
| which the University has from time to time. This page must form part of any | |
| such copies made. | |
| \item[iii.] The ownership of certain Copyright, patents, designs, trademarks | |
| and other intellectual property (the ``Intellectual Property'') and any | |
| reproductions of copyright works in the thesis, for example graphs and | |
| tables (``Reproductions''), which may be described in this thesis, may not | |
| be owned by the author and may be owned by third parties. Such Intellectual | |
| Property and Reproductions cannot and must not be made available for use | |
| without the prior written permission of the owner(s) of the relevant | |
| Intellectual Property and/or Reproductions. | |
| \item[iv.] Further information on the conditions under which disclosure, | |
| publication and commercialisation of this thesis, the Copyright and any | |
| Intellectual Property and/or Reproductions described in it may take place | |
| is available in the University IP Policy, in any relevant Thesis | |
| restriction declarations deposited in the University Library, the | |
| University Library's regulations and in The University's policy on | |
| Presentation of Theses. | |
| \begin{itemize} | |
| \item[i.] The author of this thesis (including any appendices and/or | |
| schedules to this thesis) owns certain copyright or related rights in it | |
| (the ``Copyright'') and they have given the University of Manchester certain | |
| rights to use such Copyright, including for administrative purposes. | |
| \item[ii.] Copies of this thesis, either in full or in extracts and whether | |
| in hard or electronic copy, may be made \emph{only} in accordance with the | |
| Copyright, Designs and Patents Act 1988 (as amended) and regulations issued | |
| under it or, where appropriate, in accordance with licensing agreements | |
| which the University has from time to time. This page must form part of any | |
| such copies made. | |
| \item[iii.] The ownership of certain Copyright, patents, designs, trademarks | |
| and other intellectual property (the ``Intellectual Property'') and any | |
| reproductions of copyright works in the thesis, for example graphs and | |
| tables (``Reproductions''), which may be described in this thesis, may not | |
| be owned by the author and may be owned by third parties. Such Intellectual | |
| Property and Reproductions cannot and must not be made available for use | |
| without the prior written permission of the owner(s) of the relevant | |
| Intellectual Property and/or Reproductions. | |
| \item[iv.] Further information on the conditions under which disclosure, | |
| publication and commercialisation of this thesis, the Copyright and any | |
| Intellectual Property and/or Reproductions described in it may take place | |
| is available in the University IP Policy, in any relevant Thesis | |
| restriction declarations deposited in the University Library, the | |
| University Library's regulations and in the University's policy on | |
| the Presentation of Theses. |
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
In `@docs/thesis/main.tex` around lines 103 - 128, Update the four copyright notes
in the itemize block to exactly match the mandated March 2026 University of
Manchester policy wording, including the required pronoun wording and
capitalization/article in the final note. Preserve the existing four-item
structure and do not paraphrase the official text.
Source: MCP tools
Comment on lines
+153
to
+154
| \bibliographystyle{plainnat} | ||
| \bibliography{references} |
There was a problem hiding this comment.
🎯 Functional Correctness | 🟡 Minor | ⚡ Quick win
🧩 Analysis chain
🏁 Script executed:
#!/bin/bash
set -euo pipefail
# Map the thesis file and related LaTeX setup.
git ls-files docs/thesis/main.tex docs/thesis/*.tex docs/thesis/*.cls docs/thesis/*.sty docs/thesis/*.cfg docs/thesis/*.bib
echo
echo "== main.tex outline =="
ast-grep outline docs/thesis/main.tex --view expanded || true
echo
echo "== Search for bibliography / TOC hooks =="
rg -n --hidden --glob 'docs/thesis/*' -e 'tocbibind|tableofcontents|addcontentsline|bibliography|bibliographystyle|phantomsection|hyperref|natbib|biblatex' docs/thesisRepository: memovai/mimiclaw
Length of output: 837
🏁 Script executed:
#!/bin/bash
set -euo pipefail
sed -n '1,180p' docs/thesis/main.tex | nl -ba | sed -n '1,180p'Repository: memovai/mimiclaw
Length of output: 194
🏁 Script executed:
#!/bin/bash
set -euo pipefail
echo "== top of docs/thesis/main.tex =="
awk 'NR<=120 { printf "%4d %s\n", NR, $0 }' docs/thesis/main.tex
echo
echo "== bibliography section =="
awk 'NR>=140 && NR<=160 { printf "%4d %s\n", NR, $0 }' docs/thesis/main.texRepository: memovai/mimiclaw
Length of output: 7032
Add the bibliography to the table of contents. The report/natbib bibliography here is unnumbered, so it won’t appear in the TOC unless you add a TOC-aware hook/package such as tocbibind or an equivalent hyperlink-safe \addcontentsline wrapper.
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
In `@docs/thesis/main.tex` around lines 153 - 154, Update the bibliography setup
using \bibliographystyle and \bibliography so the unnumbered bibliography is
added to the table of contents. Use the project’s existing TOC-aware package or
add an equivalent hyperlink-safe \addcontentsline wrapper, preserving the
current natbib bibliography behavior.
Source: MCP tools
|
|
||
| @inproceedings{yi2025ecoagent, | ||
| title = {{EcoAgent}: An Efficient Device-Cloud Collaborative Multi-Agent Framework for Mobile Automation}, | ||
| author = {Yi, Biao and Hu, Xavier and Chen, Yurun and Zhang, Shengyu and Yang, Hongxia and Wu, Fan}, |
There was a problem hiding this comment.
🗄️ Data Integrity & Integration | 🟡 Minor | ⚡ Quick win
Normalize author metadata against canonical records.
These entries currently produce incorrect or placeholder author attributions in the compiled thesis. (ojs.aaai.org)
docs/thesis/references.bib#L13-L13: useXueyu Hufor the AAAI proceedings entry.docs/thesis/references.bib#L43-L44: replaceAnonymouswithJabbarvaziri, Faramarz.docs/thesis/references.bib#L180-L180: replace the placeholder withHu, Yuanzhe and Wang, Yu and McAuley, Julian.
📍 Affects 1 file
docs/thesis/references.bib#L13-L13(this comment)docs/thesis/references.bib#L43-L44docs/thesis/references.bib#L180-L180
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
In `@docs/thesis/references.bib` at line 13, Normalize the author metadata in
docs/thesis/references.bib at lines 13-13, 43-44, and 180-180: use Xueyu Hu for
the AAAI proceedings entry, replace Anonymous with Jabbarvaziri, Faramarz, and
replace the remaining placeholder with Hu, Yuanzhe, Wang, Yu, and McAuley,
Julian.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary by CodeRabbit