* feat(anthropic): add Claude Opus 5 model support
Wire claude-opus-5 through the Claude 5 contract seams: adaptive-by-default
thinking with the full low..max effort range, default-sampling and prefill
stripping, streaming refusal contract, model-bound thinking replay, and 1M/128k
catalog metadata across anthropic, claude-cli, Vertex, Bedrock, and Mantle.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* fix(anthropic-vertex): normalize explicit Vertex Opus 5 model rows
Extend normalizeAnthropicVertexResolvedModel so user-configured Vertex Opus 5
rows regain reasoning, image input, 1M/128k limits, and the native thinking
map; update the live-model priority expectation and regenerate docs_map.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
* feat(sdk): always persist media facts and ship facts-first replacements for legacy Media* surfaces
PR 1 of the media legacy retirement program (audit-frozen, 4 PRs).
- Every media-bearing user turn now persists normalized __openclaw.media
facts unconditionally while continuing to emit the legacy top-level
Media* projection byte-identically (dual-write bridge; the conditional
shouldPersistStructuredMediaEntries gate now always includes media).
- New replacement APIs, shipped before any removal: typed hook media
facts (media[], originalMedia[], mediaStagingPending) on message
events; {{AttachmentPath}}/{{AttachmentUrl}}/{{AttachmentContentType}}/
{{AttachmentDir}}/{{AttachmentIndex}} template variables; focused
openclaw/plugin-sdk/media-local-roots subpath split out of the
deprecated agent-media-payload facade.
- Every legacy surface carries @deprecated naming its replacement, under
one named compatibility record media-legacy-projection with the
operator-approved removeAfter 2026-10-01 (two release trains; deletion
additionally gates on a clean published-plugin artifact sweep).
- Generic transcript append invariant documented; SDK migration, hooks,
and configuration docs updated to the facts-first path.
Writer golden matrix proves legacy bytes and model prompt bytes are
unchanged while nested facts become unconditional. 2,189 broad media
tests green; SDK api-baseline regenerated on fresh-env Testbox.
* feat(sdk): register media-local-roots subpath exports and deprecation metadata
Completes PR 1: package export map for openclaw/plugin-sdk/media-local-roots
plus the deprecated-subpath inventory and doc metadata entries for the
media-legacy-projection record.
* chore(sdk): track media-local-roots entrypoint and deprecated-export budgets
* fix(sdk): keep deprecated MSTeams buildMediaPayload re-export through the compat window
Deleting shipped runtime-api re-exports belongs to retirement PR 4 after
the media-legacy-projection window; PR 1 only deprecates. Also formats
the migration-guide schedule table.
* docs: regenerate docs map for media migration additions
* feat(approvals): emit bold headers and labels in approval prompts
Approval prompts carried plain-text labels, so iMessage showed no formatting
even though its send path now translates markdown into attributed-body ranges
(the markdown-core profile refactor, #113002). Emit bold on the headers and
field labels so channels that render markdown show formatted approval text:
iMessage into native ranges, other markdown channels into their native bold,
and channels that downgrade drop the markers cleanly.
Closes#85954.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Q7d86Ww4vJwxJwY4z1AVx6
* test(approvals): update prompt-text assertions for bold labels
* feat(approvals): bold the auto-review rationale in reaction prompts
The rationale is the reason for the interruption, so it should stand out.
Generated text, but the reaction-runtime renderers parse to an IR that
tolerates stray markers, so a rationale containing a lone marker degrades
gracefully rather than breaking the emphasis span.
* fix(approvals): preserve reaction binding and Signal rendering for bold prompts
Codex + local ClawSweeper caught that bolding the prompt headers/labels broke
downstream consumers of the visible approval text:
- Reaction/tapback binding on iMessage, Signal, and WhatsApp anchors on the
plain `Exec approval required` / `ID:` format. Strip `**` markers in each
channel's binding parser before matching, so binding still correlates the
delivered prompt. Adds an iMessage bold-format binding regression test.
- Signal sent the approval payload with textMode "plain", so the markers would
reach users literally. Switch Signal's approval sends to markdown mode;
markdownToSignalText renders the headers as native bold. WhatsApp already
renders markdown by default; iMessage renders via extractMarkdownFormatRuns.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Q7d86Ww4vJwxJwY4z1AVx6
* style(approvals): oxfmt the touched approval files
---------
Co-authored-by: Omar Shahine <10343873+omarshahine@users.noreply.github.com>
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The memory plugin's LanceDB path no longer regex-extracts or rewrites
[media attached:] marker text. Media-note lines are classified as
presentation-only and dropped wholesale during capture/query ingestion
(bracket-safe classifier: filenames may contain ']'); captions and
non-media text are preserved; legacy inline tokens stay inert escaped text.
Recall-side neutralization is deleted deliberately: image discovery is
scoped to the current-turn imagePrompt, so recalled memory text (markers
or bare path-shaped tokens) never reaches attachment scanning. Regressions
prove that boundary at the embedded, CLI, and plugin-harness entry points.
