|
| 1 | +# Post-Mortem Report: Shepherd Task Batch [#2144](https://github.com/github/copilot-sdk/issues/2144), [#2145](https://github.com/github/copilot-sdk/issues/2145), [#2146](https://github.com/github/copilot-sdk/issues/2146), [#2147](https://github.com/github/copilot-sdk/issues/2147), [#2148](https://github.com/github/copilot-sdk/issues/2148), [#2149](https://github.com/github/copilot-sdk/issues/2149), [#2150](https://github.com/github/copilot-sdk/issues/2150), [#2151](https://github.com/github/copilot-sdk/issues/2151) |
| 2 | + |
| 3 | +**Campaign:** Java embed Rust CLI runtime shepherding (`shepherd-tasks-20260730-0156`) |
| 4 | +**Report generated:** 2026-07-30 02:00 (local) |
| 5 | +**Period covered:** 2026-07-30 01:56:50 -> 2026-07-30 02:00:02 (local) |
| 6 | +**Script exit code:** 0 |
| 7 | +**Base branch:** `edburns/1917-java-embed-rust-cli-runtime-dd-3039145` |
| 8 | + |
| 9 | +## Table of Contents |
| 10 | + |
| 11 | +- [Section 1: Executive Summary](#section-1-executive-summary) |
| 12 | +- [Section 2: System Architecture](#section-2-system-architecture) |
| 13 | + - [2.1 Copilot Coding Agent (CCA)](#21-copilot-coding-agent-cca) |
| 14 | + - [2.2 Copilot Code Review Agent (CCRA)](#22-copilot-code-review-agent-ccra) |
| 15 | + - [2.3 Local Copilot CLI (Shepherd)](#23-local-copilot-cli-shepherd) |
| 16 | +- [Section 3: Per-Task Metrics](#section-3-per-task-metrics) |
| 17 | + - [Issue Legend](#issue-legend) |
| 18 | + - [3.1 — Issue #2144 / PR #2152](#31--issue-2144--pr-2152) |
| 19 | + - [3.2 — Issue #2145](#32--issue-2145) |
| 20 | + - [3.3 — Issue #2146](#33--issue-2146) |
| 21 | + - [3.4 — Issue #2147](#34--issue-2147) |
| 22 | + - [3.5 — Issue #2148](#35--issue-2148) |
| 23 | + - [3.6 — Issue #2149](#36--issue-2149) |
| 24 | + - [3.7 — Issue #2150](#37--issue-2150) |
| 25 | + - [3.8 — Issue #2151](#38--issue-2151) |
| 26 | +- [Section 4: Aggregate Statistics](#section-4-aggregate-statistics) |
| 27 | +- [Section 5: AI Credits and Token Usage](#section-5-ai-credits-and-token-usage) |
| 28 | +- [Section 6: Wall-Clock Timeline](#section-6-wall-clock-timeline) |
| 29 | +- [Section 7: Failure Analysis](#section-7-failure-analysis) |
| 30 | +- [Section 8: Observations and Recommendations](#section-8-observations-and-recommendations) |
| 31 | + |
| 32 | +--- |
| 33 | + |
| 34 | +## Section 1: Executive Summary |
| 35 | + |
| 36 | +The run completed one task end-to-end and then stopped, yielding partial campaign completion: [#2144](https://github.com/github/copilot-sdk/issues/2144) was merged via [#2152](https://github.com/github/copilot-sdk/pull/2152), while [#2145](https://github.com/github/copilot-sdk/issues/2145)-[#2151](https://github.com/github/copilot-sdk/issues/2151) had no phase artifacts in this run directory. |
| 37 | + |
| 38 | +| Metric | Value | |
| 39 | +|--------|-------| |
| 40 | +| Target tasks | 8 ([#2144](https://github.com/github/copilot-sdk/issues/2144)-[#2151](https://github.com/github/copilot-sdk/issues/2151)) | |
| 41 | +| Tasks with completed phase 1 + phase 2 | 1/8 (12.5%) | |
| 42 | +| Merged PRs | 1 ([#2152](https://github.com/github/copilot-sdk/pull/2152)) | |
| 43 | +| Campaign elapsed (first start to final event) | ~3m 12s | |
| 44 | +| Phase 1 duration for completed task | 1m 13s | |
| 45 | +| Phase 2 duration for completed task | 1m 54s | |
| 46 | +| Total per-task duration ([#2144](https://github.com/github/copilot-sdk/issues/2144)) | 3m 07s | |
| 47 | +| Copilot review rounds observed (phase 2) | 1 | |
| 48 | +| Copilot review comments generated | 0 | |
| 49 | +| Idle markers in JSON logs | 2 (one in each phase session) | |
| 50 | + |
| 51 | +--- |
| 52 | + |
| 53 | +## Section 2: System Architecture |
| 54 | + |
| 55 | +### 2.1 Copilot Coding Agent (CCA) |
| 56 | + |
| 57 | +CCA handled issue assignment-to-PR generation and produced draft PR [#2152](https://github.com/github/copilot-sdk/pull/2152) for task [#2144](https://github.com/github/copilot-sdk/issues/2144), targeting `edburns/1917-java-embed-rust-cli-runtime-dd-3039145`. |
| 58 | + |
| 59 | +### 2.2 Copilot Code Review Agent (CCRA) |
| 60 | + |
| 61 | +CCRA (`copilot-pull-request-reviewer[bot]`) posted a review entry indicating it was unable to review files for [#2152](https://github.com/github/copilot-sdk/pull/2152), resulting in zero actionable inline comments for the phase-2 fix loop. |
| 62 | + |
| 63 | +### 2.3 Local Copilot CLI (Shepherd) |
| 64 | + |
| 65 | +Local Copilot CLI orchestrated both skills (`from-assignment-to-ready`, then `from-ready-to-merged-to-base`), approved workflow gates, validated checks, adapted merge strategy (squash after merge-commit rejection), merged [#2152](https://github.com/github/copilot-sdk/pull/2152), and closed [#2144](https://github.com/github/copilot-sdk/issues/2144). |
| 66 | + |
| 67 | +--- |
| 68 | + |
| 69 | +## Section 3: Per-Task Metrics |
| 70 | + |
| 71 | +### Issue Legend |
| 72 | + |
| 73 | +| Issue | PR | Notes | |
| 74 | +|------:|---:|-------| |
| 75 | +| [#2144](https://github.com/github/copilot-sdk/issues/2144) | [#2152](https://github.com/github/copilot-sdk/pull/2152) | Completed in this run; merged and issue closed | |
| 76 | +| [#2145](https://github.com/github/copilot-sdk/issues/2145) | — | No `phase1-task-*2145*` or `phase2-task-*2145*` artifact in this run directory | |
| 77 | +| [#2146](https://github.com/github/copilot-sdk/issues/2146) | — | No artifacts in this run directory | |
| 78 | +| [#2147](https://github.com/github/copilot-sdk/issues/2147) | — | No artifacts in this run directory | |
| 79 | +| [#2148](https://github.com/github/copilot-sdk/issues/2148) | — | No artifacts in this run directory | |
| 80 | +| [#2149](https://github.com/github/copilot-sdk/issues/2149) | — | No artifacts in this run directory | |
| 81 | +| [#2150](https://github.com/github/copilot-sdk/issues/2150) | — | No artifacts in this run directory | |
| 82 | +| [#2151](https://github.com/github/copilot-sdk/issues/2151) | — | No artifacts in this run directory | |
| 83 | + |
| 84 | +### 3.1 — Issue [#2144](https://github.com/github/copilot-sdk/issues/2144) / PR [#2152](https://github.com/github/copilot-sdk/pull/2152) |
| 85 | + |
| 86 | +| Metric | Value | |
| 87 | +|--------|-------| |
| 88 | +| Phase 1 session | `17b7a2f5-bf46-4ba8-b9a0-6f990aa14e48` | |
| 89 | +| Phase 1 start / duration | 01:56:50 / 1m 13s | |
| 90 | +| Phase 2 session | `1402e0e2-bfa1-4118-8169-1665f1f04b04` | |
| 91 | +| Phase 2 start / duration | 01:58:08 / 1m 54s | |
| 92 | +| Total duration | 3m 07s | |
| 93 | +| Review rounds | 1 | |
| 94 | +| Comments generated | 0 (`"Copilot wasn't able to review any files"`) | |
| 95 | +| Outcome | merged ([#2152](https://github.com/github/copilot-sdk/pull/2152)); issue closed ([#2144](https://github.com/github/copilot-sdk/issues/2144)) | |
| 96 | + |
| 97 | +### 3.2 — Issue [#2145](https://github.com/github/copilot-sdk/issues/2145) |
| 98 | + |
| 99 | +No phase artifacts were recorded for this issue in `shepherd-tasks-20260730-0156`. |
| 100 | + |
| 101 | +### 3.3 — Issue [#2146](https://github.com/github/copilot-sdk/issues/2146) |
| 102 | + |
| 103 | +No phase artifacts were recorded for this issue in `shepherd-tasks-20260730-0156`. |
| 104 | + |
| 105 | +### 3.4 — Issue [#2147](https://github.com/github/copilot-sdk/issues/2147) |
| 106 | + |
| 107 | +No phase artifacts were recorded for this issue in `shepherd-tasks-20260730-0156`. |
| 108 | + |
| 109 | +### 3.5 — Issue [#2148](https://github.com/github/copilot-sdk/issues/2148) |
| 110 | + |
| 111 | +No phase artifacts were recorded for this issue in `shepherd-tasks-20260730-0156`. |
| 112 | + |
| 113 | +### 3.6 — Issue [#2149](https://github.com/github/copilot-sdk/issues/2149) |
| 114 | + |
| 115 | +No phase artifacts were recorded for this issue in `shepherd-tasks-20260730-0156`. |
| 116 | + |
| 117 | +### 3.7 — Issue [#2150](https://github.com/github/copilot-sdk/issues/2150) |
| 118 | + |
| 119 | +No phase artifacts were recorded for this issue in `shepherd-tasks-20260730-0156`. |
| 120 | + |
| 121 | +### 3.8 — Issue [#2151](https://github.com/github/copilot-sdk/issues/2151) |
| 122 | + |
| 123 | +No phase artifacts were recorded for this issue in `shepherd-tasks-20260730-0156`. |
| 124 | + |
| 125 | +--- |
| 126 | + |
| 127 | +## Section 4: Aggregate Statistics |
| 128 | + |
| 129 | +| Metric | Value | |
| 130 | +|--------|-------| |
| 131 | +| Tasks targeted | 8 | |
| 132 | +| Tasks with observed execution artifacts | 1 | |
| 133 | +| PRs merged | 1 ([#2152](https://github.com/github/copilot-sdk/pull/2152)) | |
| 134 | +| Issues closed | 1 ([#2144](https://github.com/github/copilot-sdk/issues/2144)) | |
| 135 | +| Success rate (target issues closed in this run) | 12.5% | |
| 136 | +| Total phase sessions exported | 2 (phase 1 + phase 2 for [#2144](https://github.com/github/copilot-sdk/issues/2144)) | |
| 137 | +| Total CCRA rounds | 1 | |
| 138 | +| Total CCRA comments | 0 | |
| 139 | +| Avg comments/round | 0.00 | |
| 140 | +| Idle markers | 2 | |
| 141 | + |
| 142 | +Convergence signal for the completed task was strong (single review round, zero fixes required), but campaign-level throughput was low because only one of eight issues progressed in this run directory. |
| 143 | + |
| 144 | +--- |
| 145 | + |
| 146 | +## Section 5: AI Credits and Token Usage |
| 147 | + |
| 148 | +| Scope | Input tokens | Output tokens | |
| 149 | +|-------|--------------|---------------| |
| 150 | +| Phase 1 JSON (`phase1-task-20260730-0156-2144.json`) | 0 | 2,934 | |
| 151 | +| Phase 2 JSON (`phase2-task-20260730-0158-2144.json`) | 0 | 3,629 | |
| 152 | +| Aggregate for run directory phase JSONs | 0 | 6,563 | |
| 153 | + |
| 154 | +No explicit CCA/CCRA billing-credit totals were present in local artifacts. Token values above are measured from local Copilot CLI JSON event logs (`assistant.message.outputTokens`/`inputTokens` fields when present). |
| 155 | + |
| 156 | +--- |
| 157 | + |
| 158 | +## Section 6: Wall-Clock Timeline |
| 159 | + |
| 160 | +| Time (local) | Event | Evidence | |
| 161 | +|--------------|-------|----------| |
| 162 | +| 01:56:50 | Phase 1 started for [#2144](https://github.com/github/copilot-sdk/issues/2144) | `phase1-task-20260730-0156-2144.md` header | |
| 163 | +| 01:58:03-01:58:04 | Phase 1 completed; [#2152](https://github.com/github/copilot-sdk/pull/2152) ready for review handoff | `SHEPHERD COMPLETE` in phase-1 MD | |
| 164 | +| 01:58:08 | Phase 2 started for [#2144](https://github.com/github/copilot-sdk/issues/2144) | `phase2-task-20260730-0158-2144.md` header | |
| 165 | +| 01:58:xx | Reviewer-request API attempts failed with HTTP 422 for `copilot-pull-request-reviewer` collaborator check | phase-2 MD command output | |
| 166 | +| 01:59:xx | CCRA review result indicated no reviewable files / 0 comments | phase-2 MD review output | |
| 167 | +| 01:59:xx | Merge via `--merge` rejected; repository requires non-merge strategy | `mergePullRequest` error in phase-2 MD | |
| 168 | +| 01:59:xx | Merge retried with `--squash` succeeded for [#2152](https://github.com/github/copilot-sdk/pull/2152) | phase-2 MD merge command output | |
| 169 | +| 02:00:xx | [#2144](https://github.com/github/copilot-sdk/issues/2144) closed; phase 2 marked complete | `gh issue close` + `SHEPHERD COMPLETE` | |
| 170 | + |
| 171 | +Overall run window observed in JSON timestamps: 01:56:52 -> 02:00:02 (~3m 11s). |
| 172 | + |
| 173 | +--- |
| 174 | + |
| 175 | +## Section 7: Failure Analysis |
| 176 | + |
| 177 | +Campaign-level failure mode was **incomplete batch execution** (1/8 issues executed in this run directory), despite `SCRIPT_EXIT_CODE=0`. |
| 178 | + |
| 179 | +Observed signatures and handling: |
| 180 | + |
| 181 | +1. Reviewer request failure: |
| 182 | + - `gh api ... requested_reviewers` returned HTTP 422 (`not a collaborator`) for `copilot-pull-request-reviewer`. |
| 183 | + - Impact: explicit reviewer request path failed; process still progressed because a bot review event appeared. |
| 184 | +2. Merge strategy mismatch: |
| 185 | + - `gh pr merge ... --merge` failed with `Merge commits are not allowed on this repository`. |
| 186 | + - Impact: resolved in-run by fallback to `gh pr merge ... --squash`. |
| 187 | +3. Session termination marker: |
| 188 | + - `Operation cancelled by user` appears at end of phase-2 export after successful merge/close messages. |
| 189 | + - Impact: may indicate interactive cancellation after essential steps completed; no evidence of additional task execution afterward in this run directory. |
| 190 | + |
| 191 | +No timeout-based hard failure was observed in this run (no stalled polls; no nonzero tool execution records in phase JSON tool events). |
| 192 | + |
| 193 | +--- |
| 194 | + |
| 195 | +## Section 8: Observations and Recommendations |
| 196 | + |
| 197 | +### What worked |
| 198 | + |
| 199 | +- End-to-end path for [#2144](https://github.com/github/copilot-sdk/issues/2144) converged quickly. |
| 200 | +- CI-gate approvals and checks completed cleanly. |
| 201 | +- Merge fallback (`--squash`) recovered from repository merge-policy rejection without manual intervention. |
| 202 | + |
| 203 | +### What did not work |
| 204 | + |
| 205 | +- Batch did not advance beyond the first issue in the provided task list. |
| 206 | +- Reviewer-request behavior is inconsistent with the configured reviewer identity and collaborator permissions. |
| 207 | +- Run-level observability is fragmented when only per-task phase artifacts exist without a unified batch log. |
| 208 | + |
| 209 | +### Recommendations |
| 210 | + |
| 211 | +1. Add an explicit post-task loop checkpoint in the shepherd driver script: after closing one issue, log and assert advancement to the next target issue ([#2145](https://github.com/github/copilot-sdk/issues/2145) here). |
| 212 | +2. Normalize review-request logic to a repository-supported path (or skip explicit request when bot auto-review is known to trigger), and treat HTTP 422 as a tracked warning with deterministic fallback. |
| 213 | +3. Detect repository merge policy before merge attempt and choose `--squash` up front when merge commits are disallowed. |
| 214 | +4. Emit a batch summary artifact listing each target issue and terminal status (`merged`, `failed`, `not-started`) to prevent ambiguity when exit code is zero. |
| 215 | +5. Capture a single top-level `*job-logs.txt` for each shepherd batch in the parent directory; none were present for this run. |
0 commit comments