Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add donald-ada/workinggenius --skill reconcilegit clone --depth 1 https://github.com/donald-ada/workinggeniusWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/donald-ada/workinggenius/reconcile)<a href="https://agentmods.dev/skills/donald-ada/workinggenius/reconcile"><img src="https://agentmods.dev/badge/skills/donald-ada/workinggenius/reconcile.svg" alt="Measured on agentmods" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00043 | $0.01210 |
| Opus 5 | $0.00022 | $0.00605 |
| Sonnet 5 | $0.00009 | $0.00242 |
| Haiku 4.5 | $0.00004 | $0.00121 |
Grade A, and why
reconcile scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured today.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 43 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Reconcile
The errata skill catches what a session trips over. Nobody trips over the rest: .genius/DECIDED.md, CONTEXT.md, DESIGN.md, ARCHITECTURE.md, the pinned verify commands, the contracts of work still in flight. Those are read as authoritative by every fresh session precisely because nothing in the flow reads them adversarially, and each keeps being obeyed until something expensive happens.
The concept: the binding documents are claims about a repo that has moved. Go check them against it.
When to run it
The user types it, at the moments the answer changes something: before designing against settled ground, after a refactor or dependency change large enough to have moved the ground, when a session has just cited a doc that turned out wrong, or when work resumes after weeks away. Not on a schedule: a scan with nothing to find will find something, and a manufactured drift is worse than the drift it invented.
What gets checked
Every line, against the repo — never against what you remember of it. With an argument, only that slug or that doc. Two of the checks are the instrument's, read as this command was invoked — every snapshot against its ceiling, and both directions of the links; where a policy notice shows instead, run it before reporting:
python3 ${CLAUDE_SKILL_DIR}/../genius-file/measure.py snapshots
python3 ${CLAUDE_SKILL_DIR}/../genius-file/measure.py links
.genius/DECIDED.md— does each decision's constraint still exist in the code, and does its link still reach the fight that settled it?CONTEXT.md— does the code still call each term what the glossary says? A rename that never reached the glossary is the collision that skill exists to prevent, arriving late.ARCHITECTURE.mdandDESIGN.md— does the code still respect each boundary and contract, do the fitness functions still exist and run, do the shipped screens still use the token roles, and do the recorded contrast ratios still hold for the values now in the code?.genius/BACKLOG.md— has the repo since satisfied a line, and do its anchors still resolve: theFromone into a work's log, the[detail]one intoBACKLOG.log.md, the[retired]links at the foot? Both can be shown. Whether a line is still worth doing cannot, so it belongs to/triage. A line the user retires leaves by the format's route for seeds, not by deletion.- In-flight work — the contract at its current version: seams, pinned numbers, conventions; and the snapshot's Open section, holding
assumed:lines nobody came back to. A snapshot over its ceiling or carrying history is drift too, and so is a work file writing paths from the repo root instead of its own folder. ⚠ The compaction one is handed to/compact, noterrata: nothing in it is wrong, only misfiled. The two link directions go there too — a link into a log whose anchor no longer exists, and a log entry no section links any more. ⚠ They do not come back the same way: an entry nothing links is repaired there, in the snapshot section it backs; a link whose target is genuinely not in the log comes back reported, not fixed, because the repair would mean writing the log, which nothing in this flow does — whether the entry went missing or the link was always wrong is the user's call. - The
## Working Geniussection — run the verify commands. A typecheck command that no longer exists is discovered here or at the worst moment of a close-out. - The links — every log entry linked from the snapshot section it backs, every done slice pointing at an anchor that exists. The block above is this check, taken mechanically; what remains is routing each line it lists — a missing anchor reported, an unlinked entry to
/compact, a key with prose after it to the format's repair.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- today Changed · +4 lines ea3ae371660c
- 3d ago Changed · +1 lines 2cd38296874b
- 8d ago First seen · 38 lines · 43 tokens per session scan A 36657443401b
reconcile is a skill published in the GitHub repository donald-ada/workinggenius (6 stars, last pushed yesterday), licensed MIT. It adds 43 tokens to every session and 1,210 once invoked, about $0.0002 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other skills, from other repositories
self-improve
Self-improve this skills repo — surface and (safely) apply quality improvements across four parallel read-only lenses (Clarity, Leanness, Coverage, Ergonomics). Sibling of /self-audit — audit fixes rule violations, improve raises quality where no rule is broken. Local meta-skill, independent of the /task: pipeline.
self-audit
Self-audit this skills repo against CLAUDE.md invariants, the artifact contract, and README/CLAUDE.md/docs sync via three parallel read-only subagents. Local meta-skill — independent of the /task: pipeline.
grill
Interrogate a plan or decision one question at a time before capture, keeping a decision-plus-rationale ledger, then route to the right capture skill.
work
Dispatch a tracker ticket into an isolated worktree and drive it autonomously to an open pull request, then stop for review. Pass interactive to brainstorm the ticket with a human before planning it, or a sentence of editorial direction as a rider. Use when asked to work a ticket end-to-end.
batch
Dispatch a night's batch of Todo tickets — one worktree and one pull request each — and leave a morning summary. Run explicitly; never inferred.
cleanup
Clean up after a merged pull request — return to the base branch, remove the worktree, delete the branch, sweep other stale branches and worktrees, and report tracker state. Use after a merge or whenever asked what is left over.