Repository navigation
Conversation
1.10.1 warned at construction whenever PII or credential findings were set to BLOCK without redact_input=True, so scanners used only for output or tool results warned for nothing. The warning now comes the first time an input scan has such a finding, once per scanner. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Rich wrapped the output at 80 columns when stdout was not a terminal, so 'last_hash <hash> (seq N)' split across two lines and broke scripts that extract the hash for --anchor. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The keyless setup from the docs (extra={'azure_ad_token_provider': ...})
raised 'AZURE_OPENAI_API_KEY is not set' unless a key was also set, though
the request only carries the bearer token. A token provider or token in
extra now counts as the credential.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A bare function was accepted at construction and failed at the first escalation with AttributeError. It is now wrapped in CallbackApprovalHandler like ApprovalPolicy(handler=...) does; anything else raises TypeError up front. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…into release/1.11.0
…elease/1.11.0 # Conflicts: # CHANGELOG.md # docs/resources/changelog.md
… release/1.11.0 # Conflicts: # CHANGELOG.md # docs/resources/changelog.md
…into release/1.11.0 # Conflicts: # CHANGELOG.md # docs/resources/changelog.md
…' into release/1.11.0 # Conflicts: # CHANGELOG.md # docs/resources/changelog.md
An AgentProcess connects its MCP servers itself and hands the agent ready-made tools, so the agent has no _mcp_multi and describe_agent found no servers: the MCP tab stayed empty. describe_agent now also collects the MCP clients behind the agent's tools. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
![github-advanced-security[bot]](https://avatars.githubusercontent.com/in/57789?s=60&v=4){kind=small}
…col, no import cycle AutoApprovalClassifier wraps a function fallback in CallbackApprovalHandler. It imported it from promptise.approval, which imports the classifier (CodeQL: cyclic import). The class now lives in _approval_types next to ApprovalHandler; promptise.approval and promptise still export the same class. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
This branch has not been deployed
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
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.
One full release with every open change: #74, #75, #76, #77, #78 and #80, on top of main (#79).
What's in it
extra={"azure_ad_token_provider": ...}orazure_ad_tokencounts as the credential, so noAZURE_OPENAI_API_KEYis required.AutoApprovalClassifier(fallback=<function>)(fix(approval): AutoApprovalClassifier accepts a function as fallback #78): a plain or async function is wrapped likeApprovalPolicy(handler=...)does; anything else is refused with aTypeErrorup front instead of failing at the first escalation.BLOCKwarning where it applies (fix(guardrails): warn about input BLOCK when input has the finding #74): the PII/credentialBLOCKwithoutredact_input=Truewarning now comes once, when input actually has such a finding, not for every scanner used only on output or tool results.promptise audit verifykeeps lines whole when piped (fix(audit): keep promptise audit verify lines whole when piped #75), solast_hash <hash> (seq N)can be read by scripts.promptise run --tuilists the MCP servers of runtime agents (fix(run): the TUI's MCP tab lists the servers of runtime agents #80): servers and each tool's server come from the MCP clients behind the agent's tools.anthropic:claude-sonnet-5-5,gemini:gemini-3.8-flashin the docs, README andpromptise models env.Checked locally
src).ruff check,ruff format --check,mypy src(242 files, clean),mkdocs build --strict.runtime_checkableApprovalHandlerprotocol, so duck-typed handlers are still accepted; the warning keys match the scanner'spii/credentialdetector names;gemini-3.8-flashis on Google's current model list.The full suite runs here in CI (Linux, macOS, Windows). After merge: tag
v1.11.0, GitHub release, PyPI, then a smoke test of the published package.Supersedes #74, #75, #76, #77, #78, #80.
🤖 Generated with Claude Code