fix(dapr-agents): rerun vs dapr-agents 1.0.5, chaining fix, real denial demo - #41
Open
StevenMih wants to merge 2 commits into
Open
fix(dapr-agents): rerun vs dapr-agents 1.0.5, chaining fix, real denial demo#41StevenMih wants to merge 2 commits into
StevenMih wants to merge 2 commits into
Conversation
…eal denial demo Rerun of the dapr_agents adapter against the current dapr-agents release (fresh venv; no dapr-agents 1.18 exists on PyPI — that pins the Dapr core SDK, not the dapr-agents package; see the drift note in the doc for both). - capsule_emit/adapters/dapr_agents.py: add prior_capsule_id to @emitter.tool() so a tool-call fyi capsule can chain onto a preceding decide capsule (previously only record_hitl() supported chaining) — needed for a real fyi -> decide(blocked) -> fyi chain. - tests/test_dapr_agents.py: cover the new chaining path. - examples/dapr-agents-capsule/demo.py: extend the demo to a real 3-capsule chain including a genuine HITL denial (verdict=blocked, effect.status=planned), anchored + inclusion-proven + offline-verified end to end. - docs/adapters/dapr_agents.md: update Limitations (L1 partially resolved — dapr-agents >=1.0.x added a native before_tool_call/RequireApproval hook system; L2/L3/L5 confirmed still true, including in that new native flow) and record the version-naming + drift findings. - pyproject.toml: bump the dapr-agents extra floor from >=0.1 to >=1.0. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> Signed-off-by: stevenmih <stevenmih88-3@gmail.com>
…in run The committed transcript was stale relative to demo.py, which was updated (2026-07-31) to script a 3-capsule chain with a real HITL denial but never re-executed. Ran it live against anchor.agentactioncapsule.org: leaf 242 (fyi), leaf 243 (decide/REJECTED, blocked), leaf 244 (fyi escalation past the denial). All three verified offline (agent_action_capsule.verify + scitt_cose.verify_receipt) and confirmed via independent curl re-check. Root cause of the first failed run: the globally registered capsule-emit editable install pointed at a different, already-merged worktree lacking the prior_capsule_id chaining parameter added in this worktree (db08e63). Re-ran `pip install -e .` from this worktree to fix. Also investigated the reported verify.agentactioncapsule.org auto-load- fragment gap: confirmed via live test + source inspection of capsule.js that fragment auto-load is currently wired and working correctly (not a bug) — documented in the transcript with the finding and recommendation. Signed-off-by: stevenmih <stevenmih88-3@gmail.com>
This was referenced Aug 6, 2026
StevenMih
added a commit
that referenced
this pull request
Aug 6, 2026
) The adapter-author half of the #22 commission; the normative half is the Privacy Considerations section in action-state-group/agent-action-capsule#35 (merged as #41). On any conflict, the profile text wins — this page says so up front. One page: the three admission classes as a table (clear-safe handles / digest-only payloads / never-enters identity+secrets, including as digests), the binding-not-naming class test, allow-list-as-code with the ADK adapter's _SAFE_CONTEXT_ATTRS tuple as the reference shape, the MCP adapter's documented client-software-only session reach as the one deliberate exception that proves the floor, and a five-question authoring checklist as the pre-ship audit. Original PR #26 (thisjody, fork context-hygiene branch) failed the neutrality gate solely because fork PRs don't receive the NEUTRALITY_TERMS secret by design. Rebased as this same-repo branch so the gate can actually run. Author credit preserved — content is Jody's own, unchanged. Signed-off-by: stevenmih <stevenmih88-3@gmail.com> Co-authored-by: stevenmih <stevenmih88-3@gmail.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
dapr_agentsadapter demo against the currentdapr-agentsrelease (1.0.5; nodapr-agents1.18 exists — see drift note).@emitter.tool()gains an optionalprior_capsule_idkwarg so a tool-call fyi capsule can chain onto a preceding decide capsule (previously onlyrecord_hitl()supported chaining).examples/dapr-agents-capsule/demo.pyextended to a real 3-capsule chain: fyi → decide (real HITL denial, verdict=blocked) → fyi (escalation), all anchored on the live public anchor and offline-verified.docs/adapters/dapr_agents.mdLimitations section updated: dapr-agents ≥1.0.x added a nativebefore_tool_call/RequireApprovalhook system (partially resolves L1); L2/L3/L5 confirmed still true, including in that new native flow.pyproject.toml:dapr-agentsextra floor bumped>=0.1→>=1.0.Full run transcript, permalinks, raw inclusion JSON, and drift findings:
_work/dapr-demo-run/(outside this repo, in the asg workspace_work/).Test plan
pytest tests/ -q— 368 passed, 6 skipped (fresh venv,dapr-agents==1.0.5installed)ruff checkcleananchor.agentactioncapsule.org(leaves 237–239), eachverify().okandverify_receiptTrue🤖 Generated with Claude Code