Named Tool Authority AI Fluency Coverage
Source-backed workplace tools now inform top-band AI Fluency coverage without pretending they prove learner proficiency.
The previously embedded named-tool authority is now connected into the top 500/top 1,000 AI Fluency coverage receipt as source context only. It can help AINA know that roles may have source-backed tool context, while preserving explicit gaps for actual learner proficiency.
What Changed
| Surface | Result |
|---|---|
| Code | ai_fluency_capability_coverage.py now reads named_tool_source_authority_v1.jsonl. |
| Matching rule | Only rows with source-backed role or workflow context can join coverage; identity-only named tools stay out. |
| Capability layer | tool_proficiency remains proxy_context, never covered, when named-tool context is joined. |
| Metrics | Coverage reports named-tool, workflow-tool, and any-tool-context candidate counts separately. |
| Test fixture | Source-backed Salesforce joins; identity-only Google Search Console does not. |
Current Proof
| Gate | Proof |
|---|---|
| Top-band coverage | status=coverage_receipt_ready_capability_hardening_required, valid=true. |
| Named-tool join | Top 1,000 has 121 named-tool candidates; top 500 has 59. |
| Boundary check | named_tool_authority_does_not_claim_tool_proficiency=true. |
| AIN-506 and AIN-510 | AIN-506 passed; AIN-510 is promotion_ready with runtime embedding authority still false. |
| Runtime readiness and validate | production-runtime-readiness and validate passed. |
| Focused tests and ruff | The coverage/source-authority tests and focused ruff check passed. |
Files To Start From
| File | Purpose |
|---|---|
artifacts/validation/ai_fluency_top_band_capability_coverage_v1.json | Machine-readable coverage receipt with named-tool metrics. |
artifacts/reports/ai_fluency_top_band_capability_coverage_v1.md | Human-readable coverage report. |
artifacts/validation/ai_fluency_top_band_capability_coverage_v1_semantic_sample_50.jsonl | 50-row sample with named-tool status included. |
artifacts/validation/named_tool_source_authority_v1.jsonl | Source-authority rows used as the named-tool join input. |
tests/test_ai_fluency_capability_coverage.py | Regression proof for source-backed join and identity-only exclusion. |
Runtime Boundary
This checkpoint does not unlock production runtime, public runtime, real-user data, external writes, or production telemetry. It also does not turn named-tool authority into learner tool proficiency. It only makes source-backed tool context visible to AI Fluency coverage, with explicit gaps preserved.
Linear Writeback
| Issue | Comment id | What was posted |
|---|---|---|
| AIN-520 | 5d631eb4-0ca8-49fa-9907-e2faf53c7ed4 | AI Fluency top-band named-tool coverage proof and boundary. |
| AIN-527 | fe8d6182-91d6-4197-95df-d15bae3380a0 | Production-readiness checkpoint proof and next hardening lane. |
Exact Resume Commands
cd /srv/aina/aina-data-engine-room
git status --short --branch
git log -5 --oneline
uv run aina-data-engine --root /srv/aina/aina-data-engine-room ai-fluency-capability-coverage
uv run aina-data-engine --root /srv/aina/aina-data-engine-room ain-506-p0-gate
uv run aina-data-engine --root /srv/aina/aina-data-engine-room ain-510-retrieval-promotion-gate
uv run aina-data-engine --root /srv/aina/aina-data-engine-room production-runtime-readiness
uv run aina-data-engine --root /srv/aina/aina-data-engine-room validate
jq '.metrics | {top_1000_named_tool_context_candidate_count, top_500_named_tool_context_candidate_count, top_1000_any_tool_context_candidate_count, top_1000_gap_reason_counts}' artifacts/validation/ai_fluency_top_band_capability_coverage_v1.jsonNext Best Work
Use this join to harden the missing 493 top-1,000 tool-context rows with source authority, not title-only guesses. The next source families should still follow the progressive ladder: eligibility, repair queue, clean or repaired corpus, dry run, semantic QA, small live run, then larger expansion only when the source-family gates stay clean.
Start with the 493 missing tool-context rows, but keep the same rule: source-backed context first, learner proficiency only after learner-confirmed or observed evidence.