aparigraha-task
Master workflow walking the Aparigraha gates end-to-end — onboarding, dependency inventory, style detection, reuse-first authoring, surgical diff, audit. Orchestrates the 25-pragmatism category skills. No implementation logic.
#advisory
33 skills tagged advisory · risk axis.
brownfield-onboarding
First-touch protocol for a new project — read README/AGENTS.md/ADRs, find build/test/CI commands, hot zones, entry points, and test layout; emit an onboarding cheat-sheet other Aparigraha skills consume.
chesterton-fence
Before deleting or refactoring code that looks unused or weird, reconstruct why it exists from history, tests, comments, and call-graph; produce a memo plus an edge-case checklist that gates the change.
code-review
When a PR needs more than a thumbs-up — review checklists, defect categories, severity classification.
constitution
The laws of the agent's universe — Satya (truth), Dharma (safety), Ahimsa (non-destruction), Pragya (wisdom).
dependency-analysis
When you're not sure what depends on what — builds dependency graphs, detects cycles, audits for staleness and CVEs.
dependency-utility-scout
Mine the project's declared dependencies and produce a per-capability inventory of utilities the team already imports — advisory, never auto-rewrites.
domain-modeling
When the team uses different words for the same thing — creates a domain glossary, entity map, and Protected Terms.
estimation
When "how long will this take?" needs a real answer — PERT estimation, relative sizing, confidence intervals.
librarian
Finds the right skill even when you describe the problem instead of naming the tool — graduated fuzzy routing.
log-analysis
When the answer is buried in 10,000 log lines — parsing, pattern recognition, event correlation, anomaly detection.
minimal-diff
Smallest correct change that solves the problem. No drive-by formatting or opportunistic refactors; diff-size caps, reversibility checks, and commit splitting when concerns are tangled.
orchestrator
Mid-task skill injection — detects domain shifts and suggests relevant skills.
pair-programming
When two heads are better than one — driver/navigator roles, mob programming, knowledge transfer protocols.
performance-review
When the app is slow but you don't know where — bottleneck identification, complexity analysis, caching review.
pr-review-flow
Master workflow for thorough PR review. Orchestrates code-review + security-review + performance-review in parallel, aggregates verdict, applies pr-management, audits. No implementation logic.
pragmatism
Aparigraha — non-accumulation. Check before create, conform within scope, stay surgical, and validate edge cases before trusting any reuse or improvement of existing code.
pragya
Direction-seeking protocol — seek corrections, present options, let humans steer.
profiling-analysis
When something is slow and you need proof of where — analyzes logs/traces to find bottlenecks and hotspots.
requirements-elicitation
When scope is fuzzy — interviews the user for goals, constraints, and success metrics before coding starts.
resource-awareness
Monitor constraints (memory, CPU, time, rate limits), adapt behavior gracefully.
reuse-first
Before writing any new utility, scan the codebase and declared dependencies for an equivalent. Reuse only after fit and edge-case validation.
risk-assessment
Before you ship something that keeps you up at night — identifies risks, scores probability/impact, plans mitigations.
root-cause-analysis
When the bug is a symptom, not the cause — 5-whys, fault trees, git bisection, symptom-to-cause mapping.
scratchpad
Forces explicit reasoning before acting — declares task, complexity, risk, and plan so nothing starts half-thought.
style-conformance
Detect the project's existing conventions (naming, formatting, errors, logging, tests) and produce a house-style profile downstream skills consult — awareness, not enforcement.
system-design
When you need to see the big picture before building — component diagrams, data flow, trade-off analysis.
task-decomposition
When a task feels too big to start — breaks it into dependency-ordered subtasks with T-shirt sizing.
tech-debt-tracking
When tech debt is everywhere but you can only fix some — categorization, impact scoring, payoff prioritization.
test-strategy
When you need a testing philosophy, not just tests — defines the pyramid, coverage targets, and conventions.
testing-strategy
When you're unsure what to test next — coverage analysis, test type selection, boundary condition identification.
token-efficiency
Resource-aware model tier, tool, and delegation selection per cognitive mode.
two-pass-analysis
When you need metrics without false-failing the build — separates collection (Pass 1) from gating (Pass 2).