Skill Gaps & Manual Agent Asks — Session History Audit

alignment_status: review

Full-history audit of Claude Code and Codex usage (12,300 user messages, Dec 2025 – Jun 2026) answering: where are the gaps in our skills, and what is still asked of agents manually? Headline: the biggest historical manual-ask families (commit/push, phase execution) were already eliminated by /ship and the exec-loop pack; the strongest remaining gap is .env scaffolding/secrets handoff (23 recent asks, no skill exists), followed by three convention-level gaps (findings verification, fix-plan routing, research freshness).

Table of Contents

1. Overview Stats

12,300
user messages
10,208 / 2,092
Claude / Codex
3,791
sessions (3,095 / 696)
75.6%
freeform (non-command) prompts
2,747
short steering prompts (≤30 chars)
MetricClaudeCodex
Date range2025-12-13 → 2026-06-102026-01-09 → 2026-06-10
Messages10,2082,092
Sessions3,095696

Top projects by message volume

ProjectMessagesShare
experiments/metternich-engine1,32210.7%
apps/bismarck-ai/v0.3/AWS1,27110.3%
poke/monorepo6945.6%
apps/bismarck-ai/v0.3/monorepo6945.6%
web/dev/bismarck-ai/bismarck-v0.46705.4%
tools/dev/agentic-skills6215.0%
web/dev/b44233.4%
poke/dev/poke-productivity-suite3703.0%
loadoutworks (apps + web)6995.7%

Monthly trend (messages)

MonthClaudeCodex
2025-12663
2026-013,640335
2026-021,57714
2026-032,208
2026-041,1361,359
2026-05733228
2026-06 (partial)251156

2. Token & Cost Check

2.02B
Codex total tokens (757 sessions)
94%
of Codex input was cached (1.889B / 2.007B)
12.9M
Codex output tokens (4.3M reasoning)
unavailable
dollar cost (no basis)

3. Skill Command Usage (all-time)

CommandUsesCommandUses
/ship857/ship-then-plan31
/clear552/workflow28
/ship-end266/investigate21
/resume166/exec20
/usage162/expert-review16
/sync124/session-triage14
/run121/analyze-sessions10
/commit-and-push-by-feature91/pack69

Long tail: /plan-interview 58, business/research skills (icp, gtm, journey-map, positioning, monetization) in single digits each. Codex $run invocations dominate Codex skill use (360+ across skill-path variants), with $ship/$sync next — Codex adoption is concentrated on run/ship/sync.

4. Closed Gaps — Manual Asks Skills Already Eliminated

The largest manual-ask families in history died out as skills were adopted — direct evidence the skill system works.

Manual ask familyVolumeTimelineReplaced by
"update docs, commit and push" / "commit this" / "push it"~552Jan 274 → Feb 206 → Mar 72 → Apr+ 0/ship (857), /ship-end (266), /commit-and-push-by-feature (91)
"implement phase N step N" / "phase N.N @docs/…plan.md"~279Jan 137 → Feb 111 → Mar 31 → Apr+ 0/ship-then-plan, /exec, exec-loop pack
"deploy to staging"59Jan 40 → Feb 19 → Mar+ 0release-ops/deploy skill (now with deployment-history tracking); bismarck project also wound down — confounder noted
"create a directory to house our new package"44Jan–Feb → Apr+ ~0monorepo pack (scaffold, mono-*)

Verified follow-through: the user's 2026-04 ask — "update the deploy skill so that it maintains a historical record of what commits were part of what deployments"was implemented: packs/release-ops/claude/deploy/SKILL.md now declares deployment history tracking in its description and body.

Q3 (required) — Do you confirm the closed-gap reading?

5. Open Gaps — Recent Manual Asks (Apr 1 – Jun 10, 2026)

Recent window: 3,863 messages, 1,892 freeform. Counts below are exact pattern matches with real examples from history.

5.1 .env scaffolding & secrets handoff — 23 recent · 69 all-time · no skill exists

Verified by inventory scan: no env-related skill in any pack. The pattern has a stable, automatable shape: derive required variables (from .env.example or code), scaffold .env.local with stubs, open it in VS Code (WSL-aware), never read secret values back. The repeated "don't look at it" warnings show the manual flow carries a recurring secret-exposure risk.

can you create the .env.local with the necessary variable stubs and then open it in vscode so I can drop it in?
can you open the .env.local in vscode? what do I need to add?
ok done, remember don't look at .env.local
no just tell me what's in .env.example. then I can add it back into .env.local. Do not edit or view the contents of .env.local

5.2 False-positive verification of findings — 10 recent · 15 all-time

expert-review already has a built-in false-positive filter step (verified in its SKILL.md), but these asks follow ad-hoc audits, not skill runs.

can you audit the codebase and hunt for any best practice reasons to reduce lines of code in our codebase?
can you double check your findings for false positives?
can you double check these findings to ensure there are no false positives?

5.3 Findings → fix-plan handoff — 11 recent · 17 all-time

After any audit/review output, the user manually requests the fix plan.

yes create a plan to fix them all
can you create a plan to fix these?
should we not make a plan to fix the significant drift? Also the 14 items recommended?

5.4 Research data freshness — explicit recurring correction (May 2026)

for the research skills, they should be attempting to get as recent data points as possible, I've had multiple instances of very stale data.

Grep across packs: only pmf-engine and youtube-portfolio carry recency language today.

5.5 Document consistency validation — 5 recent · 12 all-time

can you go over and validate the document to ensure consistency and correctness across the document?
ok is that language consistent across all docs? create a plan to audit the docs and determine if there is anything we need to update

docs-health pack (hygiene, reconcile-dev-docs) covers this but is not in this repo's enabled_packs (verified in .agents/project.json) and likely absent in the other projects where the asks occurred. Coverage exists; adoption is the gap.

5.6 Open-in-VS-Code ergonomics — 22 recent

Heavily overlaps with 5.1 — most are .env opens. Covered if the env skill includes the WSL-aware open step already documented in CLAUDE.md.

5.7 Codex-side observations

6. Buckets Reviewed and Judged NOT Skill Gaps

BucketRecent freqWhy not a gap
Design/UI tweak feedback45 (all Claude)Iterative taste feedback on animations/layout — inherently conversational; animation-design-planner + verify/run support it
Video/recording/content59guided-walkthrough + youtube-ops packs own these flows; asks were in-flow direction, not repeated workflow shapes
Research/investigate102Covered by /investigate, code-debug pack, deep-research, autoresearch
"What's next / status"22Covered by codebase-status, afps-status, roadmap skills
Run/restart dev server6Covered by /run; low frequency

7. Ranked Recommendations

#PatternRecent freqTypeSuggested actionOwner surfaceValidation
1.env scaffolding & secrets handoff23 (69 all-time)Skillenv-setup: scaffold .env.local stubs from .env.example/code, WSL-aware VS Code open, never read secretsgitops or repo-maintenance packlayer1 contract test (no secret reads; stub file produced)
2False-positive verify after ad-hoc audits10Standing instructionFindings-verification convention: self-verify pass before presenting findingsCLAUDE.md / quality-gate contract docbenchmark smoke on an audit prompt
3Findings → fix-plan handoff11Standing instructionExtend shipping-contract routing: audits end with a plan-to-fix offerCLAUDE.md shared conventionsskill-next-step-contracts doc check
4Research data freshnessrecurring correctionStanding conventionFreshness contract bundled into research skills (current-year sources, date-stamped claims, stale-evidence flags)docs/ + upgrade scriptspot-check bundled skill text
5Doc consistency validation5Pack adoption/pack install docs-health in affected projectsdocs-health packexisting pack tests

Q4 (required) — Recommendation 1: build the env-setup skill?

Q5 (required) — Recommendations 2–4 (findings-verification, fix-plan routing, research freshness conventions)?

Q6 (required) — Recommendation 5: adopt docs-health pack where doc-validation asks recur?

8. Highest-Impact View (avoided manual prompts)

  1. Historical win (already shipped): /ship + exec-loop eliminated ~830 manual commit/push/phase prompts — the single largest automation in the dataset.
  2. env-setup skill — ~23 prompts/quarter avoided, plus removes a recurring secret-exposure risk surface.
  3. Findings-verification convention — ~10/quarter avoided; improves audit quality everywhere at once.
  4. Fix-plan routing — ~11/quarter avoided; pure convention change, zero new code.
  5. Research freshness contract — prevents repeated stale-data corrections across all research skills.

9. Evidence Matrix

ClaimEvidence (source · observation · date)InferenceConfidenceDecision impact
Manual commit/push asks are eliminatedhistory.jsonl · 552 matches Jan–Mar, 0 from Apr; /ship 857 uses starting Mar · Dec 2025–Jun 2026Drop coincides exactly with /ship adoption monthHighNo action needed on git-flow skills
Manual phase-exec asks are eliminatedhistory.jsonl · 279 matches Jan–Mar, 0 from Apr · same windowexec-loop/plan skills replaced typed phase invocationsHighNo action needed
.env handoff is an uncovered recurring workflowhistory.jsonl · 23 exact recent matches, 69 all-time; pack inventory scan shows no env skill · Apr–Jun 2026Stable repeated shape + explicit safety concern → skill-shapedHighApprove/reject building env-setup (Q4)
Ad-hoc audits lack a verification passhistory.jsonl · 10 recent "false positives" follow-ups; expert-review SKILL.md already has the filter step · Apr 2026Gap is in ad-hoc behavior, not the skill — convention-level fixMedium-HighApprove/reject convention (Q5)
Findings→fix-plan handoff is manualhistory.jsonl · 11 recent "create a plan to fix" follow-ups · Apr–Jun 2026Routing convention would absorb theseMedium-HighApprove/reject convention (Q5)
Research skills lack a freshness rulehistory.jsonl · explicit user correction May 2026 ("very stale data", "multiple instances"); grep shows recency language in only 2 skillsUser explicitly says this recurs; coverage scan confirms absenceHigh (explicitly stated by user)Approve/reject convention (Q5)
Doc-validation asks are an adoption gap, not a coverage gap.agents/project.json · docs-health absent from enabled_packs; pack contains hygiene + reconcile-dev-docs · Jun 2026Skills exist; they were not installed where askedMedium (other projects' pack state not directly inspected — inferred)Approve/reject adoption rec (Q6)
Deploy-history ask was already implementedpacks/release-ops/claude/deploy/SKILL.md · description + body declare deployment history tracking · Jun 2026April user request landedHighNone — closes the loop

Q1 (required) — Is the evidence base sufficient for these recommendations?

Q2 (required) — Could the missing context change the recommendations? (Claude token data, Codex model attribution, other projects' pack state were not parsed.)

10. Confidence & Assumption Register

ConclusionStatusWhat would change it
env-setup is the top remaining skill gapEvidence-backed (exact counts, inventory scan)An existing env skill in an uninspected location, or the asks stopping naturally
Closed-gap families stay closedEvidence-backed (3 consecutive zero months)Recurrence in future history
"Deploy to staging" closure creditProvisional — confounded by bismarck wind-downA new deploy-heavy project re-surfacing the asks
Doc-validation is an adoption gapProvisional — other repos' pack state inferred, not inspectedInspecting .agents/project.json in those repos
Codex skill adoption lags ClaudeEvidence-backed (manual git-ops 38 vs 16; command mix)Codex history showing broader $-skill use
Intent-bucket countsExact for shown patterns; buckets not mutually exhaustive (first-match keyword classification)

Alternatives considered

Source coverage gaps

Downstream implications

11. Glossary Additions Gate

Proposed for research/glossary.md (flat repo — single parent glossary). Only approved terms are appended during the confirmed-page write step.

TermDefinitionSourceCategory
manual askA recurring typed user instruction that a skill could perform end-to-end (e.g. "commit this and push")analyze-sessions skill-gaps reportworkflow
steering promptA short (≤30 char) confirmation or direction message ("y", "continue", "push it") that drives an agent between skill stepsanalyze-sessions skill-gaps reportworkflow

Q7 (required) — Term: "manual ask"

Q8 (required) — Term: "steering prompt"

12. Artifact Destination & Proposed File Changes

On approval (Stage 3), the proposed canonical writes are:

Already written this run (review-stage artifacts): this alignment page, the working packet, the prompt-history log prompts/analyze-sessions/skill-prompt-20260610-020338-skill-gaps-manual-asks.md, and the index entry.

Q9 (required) — Approve the canonical artifact destination and file-change scope above?

13. Post-Approval Route

If Recommendation 1 is approved, the natural follow-up is the skill-dev pack's builder, parameterized with the gap (owner surface: gitops or repo-maintenance pack; validation: layer1 contract test asserting no secret reads and stub-file output). skill-dev is installed per .agents/project.json.

Q10 (required) — Route after this page is confirmed?

14. Compile

Feedback-only path: select emphasize / 👎 / clarify on any section above, add notes, then compile feedback YAML here (or locally under that section) — available even with gate questions unanswered.

Final approval path: answer all 10 required questions, then compile answers.