Summary
examples/capabilities/expedition-intent-agent/ and examples/capabilities/team-readiness-agent/ (renamed from examples/agents/ in #882) are fixed-output fixtures, not genuinely input-dependent logic: their src/agent.rs writes a hardcoded static OUTPUT byte string regardless of input, verified directly by reading the source.
Why this matters
This is the exact same category of issue traverse-framework/registry's decision-log entry 35 found and fixed for that repo's own reference capabilities (fixture output that never reads its input, discovered while implementing real logic for a different capability). #882 (the agent->capability terminology rename, just merged) fixed the wording of these two examples' summary field (was "Governed WASM AI agent example...", now "Governed WASM capability example...", since there was never any actual AI/model involvement to justify the old wording either) but explicitly left the underlying fixed-output-fixture problem untouched -- that was out of scope for a naming-only PR.
Ask
Give both example fixtures real, input-dependent logic (matching the standard the registry repo now holds all 11 of its own published capabilities to), or explicitly document in their manifests/README that they are intentionally static demo fixtures, not example capability logic to learn from.
Summary
examples/capabilities/expedition-intent-agent/andexamples/capabilities/team-readiness-agent/(renamed fromexamples/agents/in #882) are fixed-output fixtures, not genuinely input-dependent logic: theirsrc/agent.rswrites a hardcodedstatic OUTPUTbyte string regardless of input, verified directly by reading the source.Why this matters
This is the exact same category of issue
traverse-framework/registry's decision-log entry 35 found and fixed for that repo's own reference capabilities (fixture output that never reads its input, discovered while implementing real logic for a different capability). #882 (the agent->capability terminology rename, just merged) fixed the wording of these two examples'summaryfield (was "Governed WASM AI agent example...", now "Governed WASM capability example...", since there was never any actual AI/model involvement to justify the old wording either) but explicitly left the underlying fixed-output-fixture problem untouched -- that was out of scope for a naming-only PR.Ask
Give both example fixtures real, input-dependent logic (matching the standard the registry repo now holds all 11 of its own published capabilities to), or explicitly document in their manifests/README that they are intentionally static demo fixtures, not example capability logic to learn from.