Skip to content

mohamedzhioua/agentpace / skills/pace-estimate

pace-estimate

Verified

Estimates a plan of coding-agent tasks as P50/P80/P90 agent-time ranges with a critical path and the human part stated separately, using the agentpace CLI. Use when the user runs /pace-estimate or asks for a timed estimate of a specific task list or plan file. Natural-language "how long will this take" questions are the agentpace skill's job.

Estimate a plan

Use when the user runs /pace-estimate or /agentpace:pace-estimate, the explicit command, with a plan file or a task list. Natural-language questions belong to the agentpace skill.

Produces P50/P80/P90 for the whole plan and each task, the critical path, effective parallelism, the human part line and every warning.

The plan to estimate: $ARGUMENTS

  1. Run agentpace --version. If it fails, show this command and stop, because installing is the user's call: uv tool install agentpace

  2. If the user named a JSON plan file, use it as given. If there is no input, ask which plan or task list to estimate: a plan file that happens to be in the folder, plan.json included, is not the answer. Otherwise break the work into tasks and write a plan to a new file, never over an existing one, with this shape:

    {
      "agent_cap": 3,
      "model": "claude-opus-5-5",
      "tasks": [
        { "id": "api", "kind": "feature" },
        { "id": "review", "kind": "review", "depends_on": ["api"] }
      ]
    }
    

    Kinds: feature, bugfix, refactor, migration, tests, docs, review, other; feature is implementation work of any kind, and research is other with risk: low. Leave risk out otherwise: in the rows it is the agent's role, not how hard the work is, so high buys nothing. Set model whenever you know it; it is the strongest split in the data.

  3. Run agentpace estimate <file>. If the conversation already holds a timeline in days, weeks or months, add --quoted "<quote>" in the words it was given. Only if the user stated their own review capacity, such as "4 PRs a day", add --review-capacity "<figure>" with their words verbatim: the figure itself, in at most 80 characters; otherwise leave the flag off and repeat their figure as theirs.

  4. Start the answer with one fenced code block, tagged text, holding the CLI's own lines copied character for character: the Plan total on the session clock: line, the Quoted: line as printed when there is one, the Idle cap: line as printed when there is one, the session-clock line and the human part line, the Review capacity: line as printed when there is one, and the whole Warnings: block, heading included, when the output has one. Nothing inside the block is reworded, shortened, merged, rounded or re-ordered; whatever you want to say about a line goes below the block. Below it, name the numbers as the CLI does, "the plan total on the session clock" or "the P80 on the session clock", never as hours of agent time, active time or effort: the session clock also counts waits between runs up to the idle cap. Never say the total or its P80 is, or covers, agent time or agent work only, not even to set it against the human part; the human part line's own words, "the agent time", stay inside the block. If the CLI refuses instead of answering, with one line starting agentpace:, relay that line as printed and stop. Then name the critical path and the effective parallelism. Do not derive a number for the human part from the total, and never turn the review capacity into a timeline. Never convert the quote, compare it with the range, say what the difference between them is, or say what the quote counts: it is calendar time, as quoted, and nothing has checked it. Never reduce the result to one number.

reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis.Stars: 469kUpdated Claude CodeCodexVerifiedInstall
diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness.Stars: 321kUpdated Claude CodeCodexVerifiedInstall
dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependenciesStars: 321kUpdated Claude CodeCodexVerifiedInstall
executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is availableStars: 321kUpdated Claude CodeCodexVerifiedInstall