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.
git clone --depth 1 https://github.com/formin/spec-kit-harnessWrote 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/commands/formin/spec-kit-harness/reference)<a href="https://agentmods.dev/commands/formin/spec-kit-harness/reference"><img src="https://agentmods.dev/badge/commands/formin/spec-kit-harness/reference/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/commands/formin/spec-kit-harness/reference"><img src="https://agentmods.dev/badge/commands/formin/spec-kit-harness/reference.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 | $0.00000 | $0.01342 |
| Opus 5.5 | $0.00000 | $0.00537 |
| Sonnet 5.5 | $0.00000 | $0.00268 |
| Haiku 4.5 | $0.00000 | $0.00134 |
Grade A, and why
reference 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 1mo 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 — 44 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Command reference
All five commands are prompt files in commands/, registered in extension.yml. Each resolves HARNESS_DIR the same way: config file → SPECKIT_HARNESS_* env vars → arguments; feature directory from the current git branch (e.g. branch 003-user-auth → specs/003-user-auth/), else the most recently modified directory under specs/, else the configured fallback (default .specify/harness/global/).
/speckit.harness.init
- Purpose: create the externalized working memory — the six state files — for the active feature.
- Arguments: optional free text = the research mission; optional
key=valuebudget overrides (e.g.searches=50 inspections=60). Both may appear together. - Reads:
.specify/extensions/harness/harness-config.yml(if present). Deliberately does not readspec.md/plan.md. - Writes: creates
HARNESS_DIR/withbudget.md,candidates.md,curated.md,evidence.md,verification.md,observations.md, each written exactly once with resolved config values. - Invariants: idempotent — if
budget.mdalready exists it overwrites nothing, reports the harness as initialized, and behaves like/speckit.harness.status; a new mission in the arguments is appended as an additional numbered mission inbudget.md. Never deletes or truncates an existing harness; never copies spec/plan content into state files (links only).
/speckit.harness.explore
- Purpose: budget-aware exploration loop with strict policy/bookkeeping separation: each iteration is one policy decision (
SEARCH/INSPECT/CURATE/STOP) followed by mandatory bookkeeping. - Arguments: the research question for the session; if empty, uses the mission from
budget.md; if both empty, asks the user and stops. - Reads: slices only — budget table + last 5 action-log rows, top
curated_sliceentries, up tocandidates_sliceopen candidates, lastobservations_sliceobservations. - Writes: all six state files — appends observations, adds/updates candidate rows, promotes findings into
curated.md+ pointer entries inevidence.md, and accounts every budgeted action inbudget.md(SEARCH→ searches,INSPECT→ inspections;CURATEcosts no budget). - Invariants: one action per iteration (no batching to dodge the ledger); evidence excerpts ≤ 25 words; stops on budget exhaustion, marginal gain (default 3 no-yield actions), mission answered, or user interrupt; never edits
spec.md/plan.md/tasks.md; if state files are missing or corrupt, stops and directs the user toinitrather than recreating state mid-loop.
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.
- 1mo ago First seen · 44 lines · 0 tokens per session scan A bae5d2be91a0
reference is a command published in the GitHub repository formin/spec-kit-harness (7 stars, last pushed 3mo ago), licensed MIT. It costs nothing until one of its globs matches a file; then it loads 1,342 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 commands, from other repositories
archive
Archive a feature specification into main project memory after merge, resolving gaps and conflicts.
release
Prepare and publish spec-driven-steroids through the tag-triggered npm release pipeline.
sddp-implement-qc-loop
Command description: Run implement and QC in a continuous loop. Argument hint: [optional: feature directory or branch name] Command category: orchestration Prerequisites: spec, plan, tasks.
sddp-plan
Command description: Create an implementation plan from the current feature specification. Argument hint: [optional: planning constraints or focus areas] Command category: feature-delivery Prerequisites: spec.
sddp-qc
Command description: Run quality control against the implemented feature. Argument hint: [optional: testing focus such as unit tests, security audit, requirements sync] Command category: feature-delivery Prerequisites: spec, plan, tasks, implementation:complete.
sddp-analyze
Command description: Audit spec, plan, and tasks for consistency and quality. Argument hint: [optional: analysis focus or remediation request] Command category: feature-delivery Prerequisites: spec, plan, tasks.