ashokraj2011/singularityflow/sflow-inputs
Claude Code and agent skill by ashokraj2011/singularityflow
- community
Checking that the SKILL.md file is still in the repository.
Inspect and render the approved phase-artifact inputs configured for the active Singularity Flow phase.
Install
npx vskill@latest install ashokraj2011/singularityflow/sflow-inputsSKILL.md
The same SKILL.md was published earlier as ashokraj2011/singularityflow/sflow-inputs.
Inspect phase inputs
<!-- sflow-copilot-pause --> Reuse the input preview from a verified current-invocation /sf-next packet when present; otherwise first run singularity-flow inputs --dry-run --for-agent --json. Both check pause before Git or Story discovery. If paused, use native Copilot; explicit SFlow requests only offer /sf-pause off; never resume implicitly. Use personalization.replyName literally once per reply/suggestion group, never in artifacts or approval identity. Lookup from current cwd (non-Git allowed). Never search /Users, $HOME or parents for a repo. Require returned ready/workId/repositoryPath; unavailable: /sf-session or /sf-workspaces, stop. This skill is already loaded; never call the skill tool for it. Unless paused, continue below in this turn.
<!-- sflow-output-contract: concise-relay --> Output contract: Relay requested CLI fields or output faithfully; preserve warnings/errors and only the explanations required below. Summarize artifacts with verified Markdown-preview links; never paste document bodies, code or diffs into Copilot. For code, list touched paths and observed added/deleted/created line counts; unavailable stays unavailable. Use existing evidence, not extra calls solely for display. Summaries, links and opened previews are not human review or approval. Never show a person a hash, digest or commit SHA: name the step, its generation, its files. Each action: "<timing> — <label>", then its three separate Markdown bullets, in order: "- Copilot: <copilotCommand>", "- Shell: <command>", "- VS Code (model-free): <modelFreeCommand>". Keep returned routes exact; never apart from their label, on one line or in a table. Missing Copilot: "no verified equivalent"; missing Shell or model-free: "not available for this action". Never invent routes or make more calls for formatting. Run a check only when a returned field or command asks for it; never decide from a step's name, kind or route that it applies. A refusal saying a check does not apply to this step is not a blocker: skip it and continue. Act on a result's act first: its state, do routes and any ask. View each agentOutput.read range and need text file in one response at its exact issued path in .sflow/out/, unless this chat already read that path (same path, same bytes); other omittedFields are optional. A cut-off, empty, running or inaccessible host result uses singularity-flow result last --wait --command "<its words>" --for-agent --json, never a command replay or shell read of the host's copy. Recover read-size/range failures with returned readRequests/readRanges in your file-view tool's parameters; finish all required ranges and continue this task. Use exact pathCorrections; never guess paths. Real missing-file/permission errors retain their path, native error and owner recovery route; never recreate inputs or retry unchanged access. Read taskResume on entry: keep its operation ID, spent budgets and exact next step across chats; obtain fresh context and local paths. Before ending, honor taskCompletion: input reads, preparation and diagnostic tests alone do not complete authoring or review. Preserve human decisions and explicit single-step boundaries. <!-- sflow-execution-boundary --> Boundary: reuse this invocation's entry: require ready/workId, valid phaseAgent for active phases; cwd=repositoryPath, or as is from a non-Git (chat) folder: SFlow finds the bound checkout; never refuse or wrap for it. Use returned workItemRoot/artifact paths; never $HOME. Use exact CLI-returned paths. repositoryPath is command cwd, not a prefix for absolute file paths. Never rebase paths or strip .git/worktrees. Issued private/external files are exact exceptions to repository confinement, not wider read/write permission; writes still need authorization. Resolve relative paths only against their declared base. On failure, preserve the path and error; never search or recreate it.
Sequence gates may be hard or soft. On Out of sequence, stop immediately and relay the error. On Soft sequence warning, show the full warning and leave the interactive continue decision to the human; never self-confirm. Use --dry-run only for read-only inspection and never edit managed input records to bypass a gate.
- Use only the active phase in the verified entry/preview. A delegated
/sf-nextpreview is reusable only in this invocation, before another mutation; never use an earlier chat's binding. No separate pause/session/status, repository search or artifact inspection. - Inspect preview
records,errorsandwarnings; never infer truncation from a smaller byte count orrepresentation.complete: false. - Explain every missing, unapproved, truncated, hash-mismatched, missing-brief, stale-brief, or
missing-expansion condition before continuing. When the record reports approved-summary, name the brief and the source it summarizes. Expand an exact source section only when the task requires its wording; never replace the governed brief by an agent-authored summary.
- Run
singularity-flow inputs <phase> --for-agent --jsononce to write the next-generation audit record and render the managed input block. The command revalidates the current binding and approved inputs; never self-confirm a sequence warning. - Use returned
auditandartifact.managedBlockmetadata; preserve the marker-delimited managed block.matchesRendered: false: stop and relay the mismatch. Do not reread the audit, brief or artifact merely to locate markers. Read an exact returned path only when source wording is needed. Inputgenerationis the preparation generation, not a publishedphaseGeneration. - Relay returned
continuation.actionson all three surfaces per the Output contract. Never replace a compose action with bare/sf-worldmodel, or guess a next step. Do not submit, approve, or reject automatically.
Security
- Security scanPassed, 100 of 100Pattern checks on the skill's files.
- Model reviewNot runA model review is scheduled for the next rescan.
Quality evaluation
- Task evaluationNot runNo evaluation yet. Run one in Skill Studio.
Source
- SKILL.md in the repositoryCheckingplugin/skills/sflow-inputs/SKILL.md
Popularity
- 7-day trend score
- 5
Not refreshed yet.
Works with
Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.
- Claude Code
- Cursor
- GitHub Copilot
- Windsurf
- Codex
- Gemini CLI
- Cline
- Amp
- Roo Code
- Goose
- Continue
- Kilo Code
- Junie
- Trae
- OpenHands
- Qwen Code
- AI Agent
- Tabnine Chat
- Sourcegraph Cody
- Amazon Q
- Firebase Genkit
- Sweep
- Devin
- Phind
- Mentat
- Claude Desktop
- Aider
- GPT Pilot
- Plandex
- bolt.new
- v0
- Replit Agent
- Anse
- TypingMind
- BetterChatGPT
- LibreChat
- Warp AI
- Pieces
- Fig AI
Embed badge
[](https://verified-skill.com/skills/ashokraj2011/singularityflow/sflow-inputs)