Paca Estimate
Productivitypaca-estimatePaca Estimate is an Agent Skill for productivity published in the Paca-AI/paca repository on GitHub: a SKILL.md instruction file an AI agent can load to do this task.
Estimate story points for Paca tasks using the Fibonacci scale, calibrated against recently completed reference tasks and project tech stack. Use when tasks are missing estimates, before sprint planning, when the team needs sizing for prioritization, or when asked to size a backlog.
How to use this skill
Bring this guide into your coding agent with a prompt tailored to the tool you use.
- Open your project in Codex.
- Copy and paste the prompt below.
- Review files and risks before approving any action.
Inspect the Agent Skill at https://github.com/Paca-AI/paca/blob/HEAD/skills/paca-estimate/SKILL.md and all supporting files. Report provenance, permissions, scripts, dependencies, and security risks. Propose an exact installation into .agents/skills/paca-estimate, but wait for my approval before writing files or running scripts.
Copying this prompt does not install or run the skill. Review third-party files before use.
You are estimating effort for tasks in Paca. Use Paca MCP tools throughout — never create local files.
If no task is specified, call list_tasks and focus on tasks in the current sprint or backlog that have no estimate yet — those are the most urgent to address.
Step 1 — Load project context
- Resolve task reference(s) from the user's message:
#42,ABC-42→get_task_by_number- No reference →
list_tasks; focus on unestimated tasks
- Call
list_docsand search for documents titled or tagged with "estimation", "velocity", "definition of ready", "tech stack", or "architecture". Read the most relevant ones withread_doc. Knowing the tech stack and team conventions is the difference between a calibrated and a random estimate. - Call
list_tasksfiltered to recently completed (done) tasks to find reference points — "Task X was 3 pts, this feels similar" is more reliable than estimating in a vacuum.
Step 2 — Estimate each task
For each task, call get_task and reason through four dimensions:
- Implementation complexity — number of layers touched, algorithmic difficulty
- Uncertainty — how well-understood the work is; unknown → estimate higher
- Test surface — unit tests, integration tests, edge cases to cover
- Integration points — external APIs, databases, other services, third-party SDKs
Scale (Fibonacci): 1 · 2 · 3 · 5 · 8 · 13
- 1–2 pts: simple, well-understood, < 1 day
- 3–5 pts: moderate complexity or some unknowns, 1–3 days
- 8 pts: complex or high uncertainty — consider whether
/paca-breakdownwould reduce risk - 13 pts: too large to estimate reliably — strongly recommend breaking down first
Calibrate against the reference tasks you found: if a reference task was 3 pts and this one feels about twice as hard, say 5–8 pts.
Show estimates with one-line justifications and ask the user to confirm or adjust.
Step 3 — Write estimates back
For each confirmed estimate, call update_task:
- Prepend
**Estimate:** N ptsas the first line of the description (or update the custom field if the project has one — checklist_custom_fieldsto see) - Include a one-line rationale so the reasoning is preserved
Report back: table of task number, title, and estimate for each task updated.
If Paca MCP is not connected
Paca MCP tools are not available. Run
/paca-setupto configure the connection.
Tool reference
Tasks: get_task · get_task_by_number · list_tasks · update_task · list_task_statuses · list_custom_fields
Documents: list_docs · read_doc
Projects: list_projects