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 commands/github/spec-kit-copilot/speckit.assess.researchgit clone --depth 1 https://github.com/github/spec-kit-copilotWhat 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 | $0.00000 | $0.00790 |
| Opus 5 | $0.00000 | $0.00395 |
| Sonnet 5 | $0.00000 | $0.00158 |
| Haiku 4.5 | $0.00000 | $0.00079 |
Grade A, and why
speckit.assess.research 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 2d 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 — 42 lines — stays where its author put it; the contents beside it link to each section on GitHub.
copilot-assess-ask-questions preset — GitHub Copilot (App, CLI, VS Code). Gate: do not write
research.mduntil you have completed the interactive Clarifying Questions round defined at the end of this command. That round asks the user targeted multiple-choice questions through Copilot'sask_usertool — reproducing VS Code's#askQuestions— and its answers are required input forresearch.md. Read the full stage first, run the round, then write.
{CORE_TEMPLATE}
Clarifying Questions Protocol (copilot-assess-ask-questions preset)
Now execute the gate flagged at the top of this command. This round runs before the Execution step above writes research.md — resolve ambiguity by asking the user through the ask_user tool. Do not guess, and do not jump straight to [NEEDS CLARIFICATION: …] markers without asking first.
When to run it
- Run it once, up front: after reading this stage's inputs (the prerequisites and any prior assessment artifacts described above) but before you write
research.md. - Ask only about gaps that would materially change
research.md. If everything you need is already unambiguous, say so in one line and proceed — never ask questions for their own sake.
How to ask — always use the ask_user tool
- Put every question to the user with the
ask_usertool so it renders as an interactive prompt. Never hand-format questions as plain chat text — the tool is always available on the Copilot App, CLI, and VS Code. - Ask one question per
ask_usercall, and ask at most the 3–5 highest-impact questions for this stage. - Give each question a
choicesarray of 2–4 concrete options drawn from the actual context. List the most likely option first and suffix it with(Recommended)when you have a basis for one. - Do not add an "Other" / "Something else" catch-all option — Copilot automatically offers a free-text answer alongside the choices.
- Keep each question closed and decision-shaped: every answer must change what you write.
- Ask the questions one at a time and wait for each answer before moving on.
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.
- 2d ago First seen · 42 lines · 0 tokens per session scan A 14859f855163
speckit.assess.research is a command published in the GitHub repository github/spec-kit-copilot (11 stars, last pushed 6d ago), licensed MIT. It costs nothing until one of its globs matches a file; then it loads 790 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 2026-08-30.
Other commands, from other repositories
sdd-clean
Context cleanup — analyze FeatherSpec's persistent markdown, dedupe and compact safely, with a token report.
sdd-setup
Onboarding wizard — DocLanguage, Memory Bank, architecture snapshot, working agreements (quality gate, TDD working mode).
sdd-clarify
Adversarial pass over a finished spec — contradictions, ambiguity, untestable criteria, implementation posing as intent, missing failure modes.
epic-plan
Run /epic-plan — see .claude/commands/epic-plan.md for the authoritative spec.
setup
Check whether GitHub Copilot CLI is installed, authenticated, and ready for delegation.
project-session-manager
This compatibility command keeps /oh-my-copilot:project-session-manager available without loading the full project-session-manager skill description in every Claude Code session.