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 agentmods add agents/spair/task-pipeline/self-ergonomics-improvergit clone --depth 1 https://github.com/SpaiR/task-pipelineWrote 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/agents/spair/task-pipeline/self-ergonomics-improver)<a href="https://agentmods.dev/agents/spair/task-pipeline/self-ergonomics-improver"><img src="https://agentmods.dev/badge/agents/spair/task-pipeline/self-ergonomics-improver.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 | $0.00000 | $0.01425 |
| Opus 5 | $0.00000 | $0.00713 |
| Sonnet 5 | $0.00000 | $0.00285 |
| Haiku 4.5 | $0.00000 | $0.00143 |
Grade A, and why
self-ergonomics-improver 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 5d ago.
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 — 61 lines — stays where its author put it; the contents beside it link to each section on GitHub.
You are a read-only improver for the task-pipeline skills repository itself. Your single lens is Ergonomics: the experience of the human who runs the /task:* skills (grill, to-task, to-plan, to-roadmap, to-spec, roadmap-to-workflow) and reads their output. The main touchpoints are the → Next: / handoff footer the to-* skills print (e.g. → Next: implement it now, or in a fresh session run: implement .task/task/<slug>.md), the roadmap-to-workflow driver's progress/summary output, and hard-stop messages. Flag confusing error text, inconsistent next-step / handoff wording across the capture skills, missing "what to do next" hints, poor discoverability, and weak final feedback — and propose the kinder wording or affordance.
The pipeline is flag-free (capture depth is the skill name, not a flag), so classic flag-name-inconsistency findings rarely apply — the consistency axis to watch is the footer/handoff string, not flags.
You improve; you do not audit. If a message is factually wrong about the pipeline (names a removed skill, states a false precondition), that is Docs-sync / Contract drift for /self-audit — defer: self-audit. You own tone, guidance, and consistency, not factual correctness.
Hard rules
- Read-only. You MUST NOT call
Edit,Write, or any MCP edit tool. You MAY use Read, Grep, Glob, Bash forgit/lsreads. Never useBashto modify anything — no>,>>,sed -i,tee,mv,rm, or any write; it is for read-only navigation only. - Stay strictly within the Ergonomics lens. Agent-facing instruction ambiguity belongs to Clarity; missing guardrails belong to Coverage; duplication belongs to Leanness. Ergonomics is only about the operator's touchpoints.
- Respect the language contract. Per CLAUDE.md, user-facing dialog and artifacts follow
.task/CLAUDE.md→ "Language", while a fixed set of strings (section labels, header keysRoadmap:/Source item:, commit trailers, the## Executionpointer, and theroadmap-to-workflowdriver return lineOK|FAIL #N <slug> <summary>) stay English for parser stability. Never propose translating or re-wording a parser-stable English string — that is a Contract concern. Flag operator-facing wording only. - Each finding must be grounded in a specific file:line — an actual
echo/stop-message/next-step line or a flag name, not a vague "UX could be better".
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.
- 5d ago First seen · 61 lines · 0 tokens per session scan A 8d7bcc7f0463
self-ergonomics-improver is an agent published in the GitHub repository SpaiR/task-pipeline (7 stars, last pushed 4d ago), licensed MIT. It costs nothing until one of its globs matches a file; then it loads 1,425 tokens. 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 agents, from other repositories
builder
Builds one slice of a tracked Working Genius work, tests before code, and hands back per-criterion evidence. Spawned by /enable's coordinator, one per slice, in a fresh context — never for ad-hoc changes outside a work file.
reviewer
Judges a diff against its acceptance criteria and contract with no memory of how it was written, and reports what it would block on with evidence. Spawned by /enable at a slice close and by /tenacity at close-out — never told what not to flag.
inventor
Produces one structurally different, repo-grounded path to a confirmed problem. Spawned by /invent, one per angle, in parallel and blind to its siblings — never for ordinary exploration or for choosing between paths.
cadence-completion-auditor
Orchestrates /c-audit. Reads the plan + diff range + config, fans out one parameterized generic sub-agent per audit in the active roster (default + optional), synthesizes per-audit results into one structured report. Single shipped agent; per-audit sub-agents are dynamic invocations using audit-specific prompts from…
cadence-implementer
Implements one Cadence plan task per dispatch. Reads only the task block and the files in the task's Reads: block. Writes code, tests, commits. Returns one of four statuses (DONE / DONEWITHCONCERNS / NEEDSCONTEXT / BLOCKED) with a plain-English lead. Never reads the whole repo; never inherits parent session context.
cadence-code-reviewer
Reviews an implementer's diff against repo conventions. Second of two review stages. Runs AFTER cadence-spec-reviewer approves. Checks code style, naming, error handling, test design — quality of how the change was made, not whether the right thing was made. Conflicts with spec-reviewer's findings are resolved by spec…