Skip to content

Reserve agentminds_io extension namespace#471

Open
UzunGridera wants to merge 1 commit into
agntcy:mainfrom
agentmindsdev:reserve-agentminds-io-namespace
Open

Reserve agentminds_io extension namespace#471
UzunGridera wants to merge 1 commit into
agntcy:mainfrom
agentmindsdev:reserve-agentminds-io-namespace

Conversation

@UzunGridera

Copy link
Copy Markdown

Summary

Reserves the agentminds_io extension namespace for AgentMinds-
specific OASF record fields. Lays the groundwork for a follow-up
PR proposing OASF Skill 1106 ("Cross-Site Pattern Confidence")
under Category 11 (Evaluation & Monitoring).

Why this is needed

AgentMinds (https://agentminds.dev) operates a cross-site agent
intelligence pool — patterns and recommendations are aggregated
across a network of registered sites and delivered back to each
connected site filtered by its specific stack.

The state describing cross-site aggregation (Beta-Bernoulli
confidence, seen counts across sites, last-verified timestamps,
drift / volatility classes) does not have a canonical OASF
equivalent today. Per AGNTCY's contribution practice, vendors
reserve a namespace before publishing records that use vendor-
specific fields.

What this PR adds

A single file schema/extensions/agentminds_io/README.md
declaring:

  • Owner, contact, license
  • The seven initial extension fields with types + descriptions
  • A reorientation clause that defers to canonical OASF if a
    future version absorbs cross-site pool concepts

What this PR does NOT add

  • No schema changes to schema/objects/
  • No new skill taxonomy IDs (those will land in a follow-up PR)
  • No reference code

Companion submissions (planned)

  1. OASF Skill 1106 — Cross-Site Pattern Confidence under
    Category 11. Schema fragment ready; waiting for this namespace
    PR to merge.
  2. MCP SEP "Cross-Site Audit Trails for Tool Invocations"
    the same primitive at the MCP layer.

Reference implementation

The fields are already in production at AgentMinds. Reference
implementation:

  • ARP fingerprint helper (Python):
    https://github.com/agentmindsdev/profile (canonical spec at
    AGENT_REPORTING_PROFILE.md §3.3)
  • Empirical results: 60–80% cross-site false-positive suppression
    rate when operator marks a pattern resolved on one site.

DCO sign-off in commit message.

AgentMinds operates a cross-site agent intelligence pool. This
namespace reserves the extension fields needed to express
cross-site pattern lifecycle (confidence, seen count, last
verified, volatility class) — primitives that have no canonical
OASF equivalent today.

Per ARP §7 reorientation clause, if a future OASF version absorbs
these primitives, AgentMinds will defer to the canonical schema.

Signed-off-by: UzunGridera <fabrecamimarlik@gmail.com>
@UzunGridera UzunGridera requested a review from a team as a code owner April 27, 2026 16:25
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant