Install
$ agentstack add skill-dylanmccavitt-loom-prospect ✓ scanned · ✓ verified — works with Claude Code, Cursor, and more.
Security review
✓ PassedNo issues found. Passed automated security review. · v0.1.0 How review works →
- ✓ Prompt-injection patterns
- ✓ Secret / credential exfiltration
- ✓ Dangerous shell & filesystem operations
- ✓ Untrusted network calls
- ✓ Known-malicious package signatures
What it can access
- ✓ Network access No
- ✓ Filesystem access No
- ✓ Shell / process execution No
- ✓ Environment & secrets No
- ✓ Dynamic code execution No
From automated source analysis of v0.1.0. “Used” means the capability is present in the source — more access means more to trust, not that it’s unsafe.
About
Prospect
Scout a fresh patch and plant the first marker on it. A new idea arrives with no tracker home; prospect frames its intent just enough, then lands it as a tracker initiative or project with an idea/brief document attached — the planning system of record for everything that follows. It does not spec, decompose, or build; it opens the patch so the rest of the kit can mine it.
This skill does not create tracker initiatives, projects, or documents while being validated; during real intake it opens one home for the idea and attaches a brief.
Read the envelope first
This skill drives the tracker through its MCP tools and never hardcodes one. Before creating anything, read the repo envelope that assembler generated (the .agents/envelope/ map or the repo docs it points to) for:
- the target tracker team, and whether new ideas land as an initiative or
a project in this repo;
- the domain glossary, so the idea is named in the repo's own language;
- any naming/label conventions for new planning objects.
If no envelope exists yet, route the user to assembler to generate one before landing the idea.
Land the idea
- Frame it, briefly. Restate the idea's intent, the problem it solves, and
its rough shape in a line or two, and capture the open unknowns. Do not relitigate a decided idea or run a full interview — that depth belongs to blueprint.
- Create the tracker home. Per the envelope, call
save_initiativeor
save_project on the target team for this idea. One home per idea: reuse an existing one instead of creating a duplicate.
- Attach the brief. Call
save_documentto publish an idea/brief doc on that
home — intent, problem, rough shape, known unknowns, and explicit non-goals. No acceptance criteria, file paths, or implementation detail; that comes later.
- Return the links. Report the created initiative/project and document ids
and URLs so the next step can reference the home directly.
Hand off
- Unknowns remain (feasibility, scope, or prior art unclear) ->
blueprint
(research-spike lens) to survey before speccing.
- Ready to spec (intent and shape are clear) ->
blueprintto draft the
PRD/spec on this project.
- Asked to split into issues / make tickets -> that is
blueprint
(issue-decomposition lens), not prospect; land or locate the home first, then route to blueprint with that lens.
Invariants
- Never starts implementation. No branches, worktrees, code, or PRs — this is
pure planning intake.
- Never creates issues or sub-issues. Stamping planned work is
blueprint's
job (issue-decomposition lens); prospect only opens the home.
- Reads the
assemblerenvelope for the target team and the
initiative-vs-project choice; never hardcodes a team, project, or label.
- One home per idea. Reuse before create; do not duplicate an existing
initiative or project.
- Brief, not spec. The document captures intent and unknowns, not acceptance
criteria — the spec is blueprint's job.
Source & license
This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.
- Author: dylanmccavitt
- Source: dylanmccavitt/loom
- License: MIT
Install and usage instructions live in the source repository linked above.
Reviews
No reviews yet — be the first.
Write a review
Versions
- v0.1.0 Imported from the upstream source.