Panda Suite Docs
v1.9.2GitHub ↗
Panda Suite / Workflows and routing

Routing policy

The authoritative routing table is in Panda Lead. Its decision table handles overlaps, direct invocation, missing capabilities, changed requests, and stale plans. Its handoff contract preserves scope and authorization between activities.

Lead matches exploration, recommendation, and artifact creation to the request. Before recommending an addition, it checks what existing capabilities cover and identifies the unmet outcome. A challenged recommendation calls for reassessing its premise before proposing a replacement. No addition is a valid conclusion; an explicit artifact request still authorizes scoped creation. See the v1.1.1 refinement and verification.

Routing is a model-followed instruction policy, not a keyword classifier or a Python routing service. No deterministic runtime dispatch is claimed. The explicit table makes decisions inspectable; behavior is tested with actual tasks.

Specialists are usable directly. Lead loads metadata first, then the selected workflow, then conditional references. It does not preload every skill, require multi-agent execution, or run a reasoning discipline between every activity.

Refactor owns bounded structural improvement through baseline, incremental edits and compatibility verification. Simplify retains general complexity assessment; Plan, Execute, Test and Review retain their focused deliverables. Direct Refactor requests can stop at analysis or planning. Peer guidance is optional, with no mandatory chain or new runtime. See the Refactor design and evaluation.

Learn owns understanding unfamiliar existing code through source tracing, adaptive teaching and optional practice. Research owns bounded evidence questions; Write owns requested engineering documents; Debugger owns causal failures. A tiny explanation stays direct unless a specialist is explicitly selected. Learning does not authorize application edits or execution. See Learn design and trials.

Craft is an optional explicitly requested composition workflow for selected Panda or external specialists. Lead retains general workflow selection and coordination; Craft focuses on their shared outcome contract, handoffs and acceptance evidence. Neither is a prerequisite for the other. This overlap is intentional; added behavioral benefit remains an evaluation question. See Craft design and checks.

Logo owns visual identity exploration, production, refinement, kit packaging, and audits. Brainstorm owns broader product direction. UI implementation and skill development follow their own workflows. Use Logo directly or compose it with an explicitly selected UI specialist through Craft. See the Panda Logo usage guide.

Plan defines implementation tasks and future release acceptance gates when the requested boundary warrants them. Release evaluates the actual release revision and evidence. Plan readiness describes executable scope, never release readiness; no mandatory Plan → Release chain is introduced. See the planning refinement.

Red handles explicit security scans, surface mapping, controlled lab validation and bounty preparation. Blue handles hunting, incident reconstruction, fix verification and current threat intelligence (intel: dated CVE, KEV, EPSS, OSV, OWASP, ATT&CK and advisory lookups correlated with project components). Security handles source/configuration auditing (including an unqualified "scan for vulnerabilities") and dependency triage inside an audit; Debugger retains ordinary technical failures. Blue never performs active testing; Red does so only under an engagement. No route requires a combined workflow. Both were split from Recon; see Recon P0.