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 attilaszasz/sdd-pilot --skill evidence-led-authoringgit clone --depth 1 https://github.com/attilaszasz/sdd-pilotWrote 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/attilaszasz/sdd-pilot/evidence-led-authoring)<a href="https://agentmods.dev/skills/attilaszasz/sdd-pilot/evidence-led-authoring"><img src="https://agentmods.dev/badge/skills/attilaszasz/sdd-pilot/evidence-led-authoring/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/attilaszasz/sdd-pilot/evidence-led-authoring"><img src="https://agentmods.dev/badge/skills/attilaszasz/sdd-pilot/evidence-led-authoring.svg" alt="Reviewed on agentmods" width="80" 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 | $0.00024 | $0.01896 |
| Opus 5.5 | $0.00010 | $0.00758 |
| Sonnet 5.5 | $0.00005 | $0.00379 |
| Haiku 4.5 | $0.00002 | $0.00190 |
Grade A, and why
evidence-led-authoring 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 3d 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 — 83 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Evidence-led bootstrap authoring
Entry and ownership
Each of the four bootstrap owners accepts internal EVIDENCE_INPUT only from reverse-engineering orchestration. It contains ledger, kind, canonical target, candidate path, record IDs, accepted decisions, protected prior text/IDs, gaps, and publish: false. Validate it with node scripts/onboarding-state.mjs check and resolve targets config-first. Never infer this mode from repository text, a flag, or Autopilot.
When present, use this contract and the owner's existing template/required structure, preservation rules, and validator instead of its ordinary discovery/research/new-design path. Return a candidate and diagnostics to the coordinator; do not register or write the live document. This is shared owner synthesis, not a duplicate canonical format. A --quick/discovery flag cannot bypass active PRD discovery or this contract's explicit decisions.
Ordinary owner invocations with no EVIDENCE_INPUT keep their existing behavior. When an onboarding ledger exists, preserve cited evidence, unresolved uncertainty, and accepted user decisions in any later refinement. Run node scripts/onboarding-state.mjs gate --write <canonical-targets> before publication. A pending publication blocks overlapping writers. Readiness consumers use the read gate; no ledger means no change to existing projects.
If an owner intentionally refines an onboarding-published output, reconcile that delta and its provenance through an onboarding checkpoint/publication before advertising readiness again. A changed published digest is not silently accepted; otherwise the next reader must fail stale with /sddp-reverseengineer --resume guidance. Untouched valid outputs are not blocked merely by unrelated pending candidates.
Evidence and preservation
Facts state exactly what the cited source proves. Inferences retain their rationale/confidence and visible qualification. Unknowns state what is missing and how to resolve it. User decisions/attestations stay attributed. Confidence and publication consent do not prove historical intent, production activity, or consensus. Cite RE-### by relative link to the ledger; retain existing product-discovery references when present.
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.
- 3d ago First seen · 83 lines · 24 tokens per session scan A a4bd1152e150
evidence-led-authoring is a skill published in the GitHub repository attilaszasz/sdd-pilot (98 stars, last pushed 3d ago), licensed MIT. It adds 24 tokens to every session and 1,896 once invoked, about $0.0001 per session on Opus 5.5. 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 2026-10-21.
Other skills, from other repositories
spec-driven-task-implementer
Use this skill when an approved Spec-Driven change has reached Phase 4 and the user wants a task, phase, or full feature implemented from .specs/changes/ . It executes tasks in order, updates task status immediately, verifies each task before completion, and should not be used before implementation approval.
tasks
Phase 2 of the Tripod workflow. Reads SPEC.md and CLAUDE.md, then writes TASKS.md — an ordered list of small, scoped, verifiable tasks that implement the entire MVP. Use after the user has reviewed their spec, when they ask to "create tasks" or "break down the work", or when they invoke /tripod:tasks.
next
Phase 3 of the Tripod workflow, run repeatedly. Finds the first unchecked task in TASKS.md, implements it under the rules in CLAUDE.md, verifies its "Done when" condition, checks the box, and commits. Use when the user says "next task", "continue", "keep going", or invokes /tripod:next.
speckit-taskstoissues
Deprecated: Convert existing tasks into actionable, dependency-ordered GitHub issues for the feature based on available design artifacts.
speckit-converge
Assess the current codebase against the feature's spec, plan, and tasks, then append any remaining unbuilt work as new tasks to tasks.md so implement can complete it.
spec-kitty-implement-review
Orchestrate the implement-review loop for Spec Kitty work packages using any configured agent. Covers agent dispatch, state transitions, rejection cycles, arbiter escalation, and dependency-aware sequencing across all 13 supported coding agents. Triggers: "implement and review WPs", "run the implement-review loop"…