← All creators

panda850819

User

personal skill pack

21 indexed · 0 Featured · 5 stars · avg score 76
Prolific

Categories

Indexed Skills (21)

AI & Automation Listed

careful

Safety gate for work on production, shared infrastructure, live harness paths such as ~/.agents, ~/.claude, or ~/.codex, or unfamiliar code where the next action could damage shared state. Invoke it before a destructive command; it pauses only the high-risk action for confirmation while reversible work continues.

5 Updated 3 days ago
panda850819
AI & Automation Listed

prototype

Decision probe for one unresolved logic, state, or visual design question that is cheaper to experience than discuss. Build a throwaway terminal driver or UI variants, capture the verdict in the brief, then stop. `ui` implements a chosen visual direction; production work remains a separate human-selected task.

5 Updated 3 days ago
panda850819
Code & Development Listed

review

Review a code diff when asked, before committing, or before a PR. Uses scoped evidence, risk lanes, bounded correction, and cold-context escalation. NOT browser QA, prepared-plan critique, or external artifact trust checks.

5 Updated 3 days ago
panda850819
Code & Development Listed

qa

Browser evidence route when UI changed, browser acceptance remains unproven, or the user asks to test a page. Requires host browser automation; maps the current artifact to acceptance criteria and stores a PR-ready handoff for `ship`. Use `review` for the diff and native tests for non-UI behavior.

5 Updated 3 days ago
panda850819
API & Backend Listed

gatekeeper

Pre-adoption trust route when an external skill, MCP server, repository, URL, package, API, SDK, or service may be installed, executed, granted access, or trusted. Classify STRIDE, follow the artifact-specific review, and return adopt, restrict, or reject. High-risk paths require human approval; rejected paths stop with refusal and evidence.

5 Updated 3 days ago
panda850819
Code & Development Listed

ship

Delivery route after coding work is complete and the user asks to ship, create a PR, or publish the branch. Run test, scope, and review gates; commit relevant files; push a non-default branch; then upsert the PR and available QA evidence. Unfinished work returns to its execution owner.

5 Updated 3 days ago
panda850819
Data & Documents Listed

improve-codebase-architecture

Explicitly requested, read-only survey for a repository area where the best architecture investment is unknown; ranks evidence-backed seams in a visual report for the user to choose before a later Grilling Session.

5 Updated 3 days ago
panda850819
Web & Frontend Listed

codebase-design

Deep-module design for choosing a small, testable interface at a clean seam. Use when the design question is a module's interface, seam, depth, or test surface, or when another skill needs these terms. NOT diff review (`review`) or requirement discovery (`grill`).

5 Updated 3 days ago
panda850819
AI & Automation Listed

handover

Bounded fresh-context handover for unfinished mechanical build units from an existing plan. A Claude Code or Codex orchestrator may start a fresh Claude or Codex worker and keeps ownership. - /handover [--agent claude|codex] [slug]: sync handoff through `scripts/verbs fresh-run`. - /handover --async [slug]: write a self-contained payload to docs/handoffs/ only; does not spawn Codex or touch git. Also owns explicit native Agent Worker or parallel read-only research with at most two depth-one workers. Use when a locked, file-scoped task benefits from fresh conversation context or when bounded read-only fan-out is requested. NOT for plan writing, closing finished work, PR/ship flow, or exploratory judgment-heavy work (pull a cross-model take with advisor instead).

5 Updated 3 days ago
panda850819
Code & Development Listed

sprint

Focused execution from a concrete outcome to SHIPPED, PAUSED, FAILED, or ABORTED_BY_USER. Adds acceptance-driven loops, bounded review, and delivery evidence beyond native coding behavior. Routes unclear requirements to grill, bugs to debug, UI work to ui, and completed delivery to ship. NOT for hypothetical or planning-only responses.

5 Updated 3 days ago
panda850819
AI & Automation Listed

harness-slim

Audit and reduce a live multi-runtime agent harness after adoption. Use for installed skill/plugin parity, cold-start context, routing overlap, available usage evidence, or human-attention load. Produces a reversible proposal; it does not mutate the harness.

5 Updated 3 days ago
panda850819
AI & Automation Listed

grill

Adversarial requirement discovery. Ask dependency-aware frontier rounds, hunting for hidden requirements / unknown unknowns. Its structured close routes spec-sized work to `to-spec`, large foggy work to `wayfinder`, and smaller work to a local brief plus executable plan. Say "quick" or "don't write files" for a chat-only log. Use for "grill me", "stress test this scope", "what am I missing", "draft a brief", structured intake, or a feature/refactor expected to touch 3+ files or add an abstraction. Skip when scope is already concrete.

5 Updated 3 days ago
panda850819
AI & Automation Listed

to-tickets

Turn one complete canonical GitHub Spec Issue into approved child Issues and dependency edges. Use for vertical-slice ticketing only. NOT discovery, Spec publication, scheduling, or implementation.

5 Updated 3 days ago
panda850819
AI & Automation Listed

wayfinder

Chart or work a decision map across sessions until the route to its destination is clear. Use when the request itself names a map, resumes one, or an `ask-boss` handoff identifies multi-session decision fog. With no map yet, run the interview and write the map here, then stop. With an existing map, take ONE unblocked entry, resolve it by type (research / grilling / prototype / task), write the decision back, and graduate the fog. A request without a named map that only needs one-session requirement discovery goes to `grill`. NOT executing a locked plan (sprint).

5 Updated 3 days ago
panda850819
Code & Development Listed

to-spec

Publish established discussion and repository evidence as one canonical GitHub Spec Issue. Use when settled requirements need a durable specification or `grill` routes a spec-sized effort here. NOT discovery or ticket decomposition.

5 Updated 3 days ago
panda850819
Code & Development Listed

advisor

Pull a decorrelated second opinion from a DIFFERENT model on a LOAD-BEARING judgment: a design fork, a pre-commit plan, an irreversible or outward-facing call. Reversible calls just decide. `--panel` = blind critics on a prepared plan. NOT code-diff review (review), NOT mechanical work for Codex (handover), NOT sharpening a fuzzy idea (grill).

5 Updated 3 days ago
panda850819
AI & Automation Listed

setup-verbs

Configure or repair a repository's Verbs issue-tracker setting. Use when setting up Verbs in a repo, when a tracker-dependent workflow cannot find its tracker, or when the existing `## verbs` block and Git remote may disagree.

5 Updated 3 days ago
panda850819
DevOps & Infrastructure Listed

decision-map

Cross-session navigation route when one destination depends on several decisions whose dependencies or evidence cannot fit the current Grilling Session. Create a named map or resolve one human-selected unblocked entry, optionally through `prototype`; update the frontier and stop before implementation.

5 Updated 3 days ago
panda850819
AI & Automation Listed

ask-boss

Route an unclear workplace request to one existing specialist by retrieving facts and resolving intent, target, audience, and minimum sufficient authority. Use for an unclear starting point, owner, reference, or next route; clear typed requests, named maps, bugs, UI work, and code review go directly to their specialist.

5 Updated 3 days ago
panda850819
Web & Frontend Listed

ui

Production UI craft route when a page or component needs a committed visual direction, or a taste complaint such as 不好看、突兀、死板、沒回饋感 needs correction. Name the direction, implement required states, and verify rendered locales. Use `prototype` while direction is open, `debug` for render failures, and `qa` for browser acceptance evidence.

5 Updated 3 days ago
panda850819
Code & Development Listed

debug

Root-cause route for an observed error, crash, regression, failing test, or intermittent behavior whose cause is still unknown. Reproduce it, name the evidence-backed cause before editing, verify the fix with a red-capable check, and scan for siblings. Use `review` for diff review and `ui` for taste complaints.

5 Updated 3 days ago
panda850819

Bio shown is the top-scored skill's repo description as a fallback — real GitHub bios land in a future update.