---
name: aoa-eval-session-mining
description: Mine .aoa session evidence for missed eval triggers, candidate regression cases, and local eval pressure only after web research and repo owner surfaces are checked, returning raw and segment refs with freshness rather than proof. Use when the user asks to study sessions for eval moments, missed triggers, repeated failures, or session-derived eval cases. Do not use for raw .aoa preservation, hook repair, search-index maintenance, memo writeback, decision records, or treating session history as reviewed proof.
license: Apache-2.0
compatibility: Designed for Codex or similar coding agents with repository file access and an interactive shell. Network access is optional and only needed when repository validation or referenced workflows require it.
metadata:
  aoa_scope: core
  aoa_status: scaffold
  aoa_invocation_mode: explicit-preferred
  aoa_source_skill_path: skills/core/engineering/aoa-eval-session-mining/SKILL.md
  aoa_source_repo: 8Dionysus/aoa-skills
  aoa_technique_dependencies: AOA-T-0075,AOA-T-0067,AOA-T-0081
  aoa_portable_profile: codex-facing-wave-3
---

# aoa-eval-session-mining

## Intent
Use this skill to mine `.aoa` session evidence for missed eval-trigger moments,
candidate regression cases, and local eval pressure after external and repo
owner surfaces have already been studied.

## Trigger boundary
Use this skill when:
- the user asks to study sessions for eval moments, missed triggers, regressions,
  or repeated failures
- repo and web research has already defined candidate trigger classes
- `.aoa` search hits can reveal real prompts, failures, validations, or
  corrections that should become eval cases
- the output should be candidate evidence with raw/segment refs and freshness

Do not use this skill when:
- the task is only `.aoa` preservation, hook repair, or search-index maintenance
- a current repository source file can answer the question without session
  mining
- session evidence would be treated as reviewed proof
- memory writeback, decision records, or checkpoint closeout are the real owner
  routes

## Inputs
- trigger classes or target failure modes from prior research
- `.aoa` search index status, raw refs, segment refs, freshness, and diagnostics
- target repositories, touched paths, validators, tests, and local eval ports
- review criteria for candidate vs accepted evidence

## Outputs
- missed-trigger examples with session id, label, segment, event, raw ref, and
  freshness
- candidate eval-case themes and owner routes
- rejected examples and nearest wrong owner routes
- local intake or design handoff when evidence is strong enough

## Procedure
1. confirm the research gates: web sources, repo-local eval surfaces, and owner
   boundaries are already mapped
2. check `.aoa` search provider status and index freshness
3. run narrow searches by trigger class, not one broad vague query
4. inspect enough surrounding segment context to avoid false positives
5. classify each hit as candidate eval trigger, candidate local need, existing
   eval apply moment, non-eval validation, or wrong owner route
6. preserve raw and segment refs; do not summarize them into authority
7. hand off accepted candidates to `aoa-eval-local-need` or `aoa-eval-design`
   only after owner-route checks

## Contracts
- `.aoa` evidence is candidate evidence until reviewed
- session search must include freshness and refs
- raw transcript evidence cannot override source-owned repo files
- mining happens after owner boundaries are known, not as a replacement for
  source archaeology

## Risks and anti-patterns
- using session snippets as proof without source verification
- mining all sessions before trigger classes are defined
- confusing `.aoa` maintenance with eval mining
- turning repeated frustration into a generic eval without a reproducible case

## Verification
- confirm index freshness and provider
- confirm each example has session, segment, event, raw ref, and freshness
- confirm candidate owner route and rejected nearest wrong owner
- confirm no raw evidence was promoted to proof
- confirm next local need/design/apply route when applicable

## Technique traceability
Manifest-backed techniques:
- AOA-T-0075 from `8Dionysus/aoa-techniques` at `1a7d146957108ecefc24219c7d56357c5a4a2c2c` using path `techniques/continuity/donor-harvest/session-donor-harvest/TECHNIQUE.md` and sections: Intent, Inputs, Outputs, Core procedure, Contracts, Risks, Validation
- AOA-T-0067 from `8Dionysus/aoa-techniques` at `1a7d146957108ecefc24219c7d56357c5a4a2c2c` using path `techniques/history/history-artifacts/transcript-linked-code-lineage/TECHNIQUE.md` and sections: Intent, Inputs, Outputs, Core procedure, Contracts, Risks, Validation
- AOA-T-0081 from `8Dionysus/aoa-techniques` at `1a7d146957108ecefc24219c7d56357c5a4a2c2c` using path `techniques/recovery/diagnosis-repair/diagnosis-from-reviewed-evidence/TECHNIQUE.md` and sections: Intent, Inputs, Outputs, Core procedure, Contracts, Risks, Validation

## Adaptation points
- `.aoa` search command and provider
- session ref format
- local intake handoff criteria