agentsclimarketplace

Celesteops review

Skill whykusanagi/celeste-ops/skills/celesteops-review

CelesteOps — local-first creator control panel: signed macOS app + MCP agent kit

Install
npx -y skills add whykusanagi/celeste-ops --skill celesteops-review

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

2 things to look at

  • no licenseNo license file was found in the repository. Code published without one is not open source by default, so using it at work is a question for whoever answers licensing questions where you are.
  • 0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Use when the user asks an agent to review specs, plans, notes, pending approvals, or content tracked in CelesteOps. Prefer the CelesteOps MCP tools when available; start from the pending-review queue or project rollups, read linked documents and tasks for context, summarize decision points, and only change review state when the user explicitly asks.

SKILL.md

3.6 KB, 705 tokens by cl100k_base, as published. Nobody here has run it

CelesteOps Review

Use this skill when CelesteOps is the source of truth for review state. The goal is to help the user understand what is waiting in CelesteOps, what each item is asking for, and what should happen next.

Quick start

  1. Confirm the session exposes CelesteOps MCP tools. Tool names may be namespaced; match by suffixes such as documents_pending_review, documents_pending_decisions, document_get, document_set_review_status, document_decision_resolve, documents_search, documents_for_entity, document_backlinks, tasks_search, tasks_list, projects_list, and pipeline_list.
  2. If CelesteOps tools are not available, inspect ../../README.md for the MCP connection shape and tell the user the skill cannot operate the review queue until the MCP is attached. Do not mutate SQLite directly as a substitute.
  3. For document review, start with documents_pending_review and documents_pending_decisions — a doc can wait on a user choice (an agent-authored decision) without being in the review-status queue.
  4. For repo-scoped review, also call projects_list and then the relevant task or document search tools.
  5. For content-pipeline review rather than document approval, start with pipeline_list.

Review rules

  • Read the full document with document_get before recommending approval or modification. document_get returns {document, comments, decisions} — read the embedded reasoning chain (agent comments + any open/resolved decisions) so you summarize the doc and its unresolved choices together.
  • Pull surrounding context with documents_for_entity, document_backlinks, tasks_search, or tasks_list when a doc references a task, repo, feature, or prior plan.
  • Treat CelesteOps as the workflow system of record. If auxiliary MCP tools such as celeste_index or celeste_code_review are available, use them only to validate claims or inspect referenced code; do not let them replace the CelesteOps review state.
  • Summarize review items in operator language: what this is, what decision is being requested, notable risks, and the recommended next action.
  • Only call document_set_review_status when the user explicitly asks to approve, mark modified, clear review state, or submit something for review. The same rule covers document_decision_resolve (picking the chosen option / leaving a resolution note) and approve-with-note: these record the user's decision, so perform them only on explicit instruction. When the user approves or marks-modified with a reason, pass that reason as the note so it is filed as a comment on the status change.
  • Avoid destructive cleanup during review. Prefer updating status or leaving notes over deleting docs or tasks.

Deep references

  • Use references/review-workflow.md for the normal decision tree and response shape.
  • Use ../../MCP.md only when you need exact tool contracts or edge-case behavior. It is the authoritative reference but much larger than the workflow guide.

If you need a narrow section from MCP.md, search it first:

rg -n "documents_pending_review|documents_pending_decisions|documents_review_changes_since|document_set_review_status|document_decision_resolve|document_get|documents_for_entity|document_backlinks|projects_list|pipeline_list|tasks_search" ../../MCP.md

Gives 0 of the 12 instructions most review quality skills give in 705 tokens

Counted across 1,048 of the 1,783 authors here whose files we hold, read 2026-08-06

  • ask questions one at a timein 82 of 1048, across 54 files
  • provide a recommended answer for each questionin 73 of 1048, across 45 files
  • explore the codebase instead of asking answerable questionsin 66 of 1048, across 37 files
  • resolve dependencies between decisions one-by-onein 42 of 1048, across 15 files
  • interview the user relentlessly about the planin 39 of 1048, across 12 files
  • order findings by severityin 29 of 1048
  • resolve each branch of the decision treein 28 of 1048, across 5 files
  • run a grilling sessionin 26 of 1048, across 5 files
  • update CONTEXT.md immediately when a term is resolvedin 26 of 1048, across 9 files
  • propose precise canonical terms for vague languagein 25 of 1048, across 6 files
  • create documentation files lazilyin 24 of 1048, across 5 files
  • use the domain-modeling skillin 22 of 1048, across 3 files

Said here and by no other author read

  • Confirm CelesteOps MCP tools are available before proceeding
  • Start from pending-review queue and pending decisions
  • Read the full document before recommending approval
  • Read embedded reasoning chain and unresolved choices
  • Pull surrounding context for referenced items
  • Treat CelesteOps as the system of record

Grouped from the skills themselves: near-identical wordings counted once, and counted by distinct author, so one author publishing three of these counts once. Length counted with cl100k_base; the agent that loads this file may tokenize it differently.

Keep looking

Skills are one crate of 328,083. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.