feat(parser): migrate OpenClaw + QClaw providers by mariusvniekerk · Pull Request #880 · kenn-io/agentsview

mariusvniekerk · 2026-06-26T02:41:41Z

Migrates the OpenClaw and QClaw providers onto the facade, preserving discovery and parse behavior.

mariusvniekerk · 2026-06-26T02:42:04Z

This change is part of the following stack:

feat(parser): provider facade core #876
- feat(parser): reusable source-set framework #877
  - feat(parser): migrate CLI/JSONL providers #878
    - feat(parser): migrate Qwen + QwenPaw providers #879
      - feat(parser): migrate OpenClaw + QClaw providers #880 ◀
        
        feat(parser): migrate opencode-family providers (opencode, kilo, mimocode) #881
        
        feat(parser): migrate bespoke providers (openhands, cursor, vibe, hermes, claude, cowork) #882
        
        feat(parser): migrate Codex + editor-extension providers #883
        
        feat(parser): migrate SQLite/container providers #884
        
        refactor(parser): require explicit provider factories + migration cleanup #885

_{Change managed by git-spice.}

roborev-ci · 2026-06-26T12:52:07Z

roborev: Combined Review (`5970de1`)

High-risk regression found in the OpenClaw/QClaw provider migration; security review found no exploitable issue.

High

Location: internal/parser/types.go:367
Problem: Removing DiscoverFunc/FindSourceFunc from OpenClaw and QClaw leaves existing hook-based callers behind. internal/ssh/resolve.go:77 will now skip these agents entirely for remote sync, and cmd/agentsview/token_use.go:91 / cmd/agentsview/token_use.go:112 can no longer resolve unsynced OpenClaw/QClaw IDs for on-demand session usage.
Fix: Either keep the legacy hooks until those callers are provider-aware, or update the SSH resolver and session ID resolver to include provider-authoritative file-based agents.

Medium

Location: internal/parser/claw_provider.go:312
Problem: The provider fingerprint no longer computes a content hash, so clawParseOutcome only sees an empty req.Fingerprint.Hash and full syncs will overwrite previously stored OpenClaw/QClaw file_hash values with NULL. The removed legacy processors always computed ComputeFileHash.
Fix: Populate SourceFingerprint.Hash in clawSourceSet.Fingerprint using the file content hash, or compute and stamp the hash during Parse.

Panel: ci_default_security | Synthesis: codex, 10s | Members: codex_default (codex/default, done, 5m44s), codex_security (codex/security, done, 2m41s) | Total: 8m35s

roborev-ci · 2026-06-26T15:58:41Z

roborev: Combined Review (`645790d`)

Medium issue found; no high or critical findings.

Medium

internal/parser/claw_provider.go:312 - OpenClaw/QClaw provider fingerprints now include only size and mtime, so clawParseOutcome does not receive a real Fingerprint.Hash during normal sync. Since the removed legacy processOpenClaw/processQClaw paths computed and stored ComputeFileHash, reparsing these sessions can clear sessions.file_hash.

Fix: Preserve legacy behavior by computing a content hash for the source, either in Fingerprint or as a fallback in clawParseOutcome, and add a provider-level test that exercises provider.Fingerprint instead of injecting a hash.

Panel: ci_default_security | Synthesis: codex, 7s | Members: codex_default (codex/default, done, 7m10s), codex_security (codex/security, done, 1m57s) | Total: 9m14s

roborev-ci · 2026-06-26T19:04:58Z

roborev: Combined Review (`08a36f5`)

Summary verdict: One medium issue needs fixing before merge.

Medium

internal/parser/claw_provider.go:312 - Provider fingerprints omit Hash, while clawParseOutcome only sets sess.File.Hash from req.Fingerprint.Hash. Real OpenClaw/QClaw syncs will write a nil file_hash and can clear hashes previously populated by the legacy ComputeFileHash path.
- Fix: Populate SourceFingerprint.Hash with the transcript SHA-256, or compute/set the hash during parse. Add a test that runs Fingerprint then Parse.

Panel: ci_default_security | Synthesis: codex, 7s | Members: codex_default (codex/default, done, 8m29s), codex_security (codex/security, done, 1m39s) | Total: 10m15s

roborev-ci · 2026-06-26T19:17:44Z

roborev: Combined Review (`9cdc39d`)

Medium findings remain in the OpenClaw/QClaw provider migration; no security-impacting issues were reported.

Medium

internal/parser/claw_provider.go:312 - Provider fingerprints only include size/mtime, so clawParseOutcome never receives a real Fingerprint.Hash. Legacy OpenClaw/QClaw processing computed and persisted file_hash, but provider sync may now overwrite it with an empty value. Compute a content hash in clawSourceSet.Fingerprint or during parse, and add an engine/provider-path test that verifies stored file_hash is non-empty.
internal/ssh/resolve.go:77 - Remote sync discovery still skips file-based agents with no DiscoverFunc. Since this diff removes those hooks from OpenClaw/QClaw, their remote directories are no longer emitted for transfer. Teach the resolver to include provider-authoritative file-based agents, or keep small legacy discovery adapters until remote sync is provider-aware.
cmd/agentsview/token_use.go:91 - Local session usage disk probing still only calls FindSourceFunc. OpenClaw/QClaw sessions present on disk but not yet in the DB now resolve as unknown, so on-demand SyncSingleSession is skipped. Add provider-backed disk probes for provider-authoritative file-based agents, or retain FindSourceFunc adapters for these agents.

Panel: ci_default_security | Synthesis: codex, 10s | Members: codex_default (codex/default, done, 6m13s), codex_security (codex/security, done, 1m29s) | Total: 7m52s

roborev-ci · 2026-06-26T21:58:13Z

roborev: Combined Review (`4f66903`)

Reviewed OpenClaw/QClaw provider migration: 2 medium issues need attention before merge.

Medium

internal/parser/types.go:367
Removing FindSourceFunc for OpenClaw/QClaw breaks session usage / token-use for unsynced Claw sessions. resolveRawSessionID still probes disk only through AgentDef.FindSourceFunc, so valid on-disk openclaw:<agent>:<session> or qclaw:<agent>:<session> IDs that are not yet in the DB are treated as unknown and skipped instead of being synced on demand.
Fix: Add a provider-backed disk probe to resolveRawSessionID for provider-authoritative agents, or keep compatibility FindSourceFunc wrappers until that caller is migrated. Add tests for unsynced OpenClaw and QClaw IDs on disk.
internal/parser/claw_provider.go:312
The new Claw provider fingerprint omits a content hash, but clawParseOutcome only preserves Session.File.Hash when req.Fingerprint.Hash is set. The removed legacy processOpenClaw / processQClaw paths always computed file_hash, so the next provider parse will write NULL and clear existing stored hashes.
Fix: Compute and populate SourceFingerprint.Hash in clawSourceSet.Fingerprint for Claw sources, matching the legacy full-file hash behavior.

Panel: ci_default_security | Synthesis: codex, 9s | Members: codex_default (codex/default, done, 6m7s), codex_security (codex/security, done, 1m4s) | Total: 7m20s

roborev-ci · 2026-06-27T13:01:16Z

roborev: Combined Review (`8eb63eb`)

Summary verdict: one medium regression should be fixed before merge; no security issues were found.

Medium

internal/parser/types.go:367, cmd/agentsview/token_use.go:91
OpenClaw/QClaw now drop FindSourceFunc, but resolveRawSessionID still resolves on-disk unsynced sessions only through FindSourceFunc. An unsynced Claw session present on disk is now treated as unknown, so session usage/token-use skips the on-demand SyncSingleSession path and reports not found.

Fix: Teach resolveRawSessionID to use provider FindSource for provider-authoritative file-based agents, or keep thin legacy FindSourceFunc shims for Claw. Add on-disk empty-DB tests for OpenClaw and QClaw raw/canonical IDs.

Panel: ci_default_security | Synthesis: codex, 8s | Members: codex_default (codex/default, done, 6m19s), codex_security (codex/security, done, 2m49s) | Total: 9m16s

roborev-ci · 2026-06-27T23:24:06Z

roborev: Combined Review (`def6e5d`)

Medium-risk issues remain around OpenClaw/QClaw provider migration; no security findings were reported.

Medium

internal/parser/types.go:368 - Removing FindSourceFunc from OpenClaw/QClaw breaks session usage / token-use on-demand resolution for sessions that exist on disk but are not yet in the DB. resolveRawSessionID still probes disk only through FindSourceFunc, so provider-owned Claw sessions return known=false and never call SyncSingleSession.

Suggested fix: Add a provider-backed disk probe to resolveRawSessionID, or keep thin FindSourceFunc compatibility wrappers until the CLI uses providers. Add tests for canonical and raw on-disk OpenClaw/QClaw IDs absent from the DB.
internal/parser/claw_provider.go:315 - Fingerprint now reads and hashes the entire Claw transcript. Provider-owned sources call Fingerprint during mtime filtering, so incremental syncs can hash every OpenClaw/QClaw file before excluding old files by mtime, regressing the legacy stat-only cutoff path.

Suggested fix: Keep Fingerprint cheap for freshness checks and compute the full content hash during Parse when populating Session.File.Hash, or add a separate cheap mtime path for provider filtering.

Panel: ci_default_security | Synthesis: codex, 8s | Members: codex_default (codex/default, done, 9m20s), codex_security (codex/security, done, 2m25s) | Total: 11m53s

roborev-ci · 2026-06-28T11:29:10Z

roborev: Combined Review (`48ae3e7`)

Medium-risk issue found: the Claw provider migration appears to break on-disk token/session usage lookup for unsynced OpenClaw/QClaw sessions.

Medium

internal/parser/types.go:367: Removing FindSourceFunc from OpenClaw/QClaw breaks resolveRawSessionID disk probes, which still only call AgentDef.FindSourceFunc in cmd/agentsview/token_use.go. Unsynced on-disk OpenClaw/QClaw sessions can no longer be recognized by token/session usage lookup, so the on-demand SyncSingleSession path is skipped and the CLI reports not found.

Fix: Teach resolveRawSessionID to use provider FindSource for provider-authoritative file agents, or keep compatibility FindSourceFunc wrappers for these agents. Add coverage for canonical and raw OpenClaw/QClaw IDs present on disk but absent from the DB.

Panel: ci_default_security | Synthesis: codex, 7s | Members: codex_default (codex/default, done, 4m16s), codex_security (codex/security, done, 2m37s) | Total: 7m0s

roborev-ci · 2026-06-28T14:16:45Z

roborev: Combined Review (`e15adbf`)

No issues found.

Panel: ci_default_security | Synthesis: codex | Members: codex_default (codex/default, done, 6m58s), codex_security (codex/security, done, 2m45s) | Total: 9m43s

roborev-ci · 2026-06-28T15:44:22Z

roborev: Combined Review (`edf4f0f`)

PR is not ready: one medium regression needs to be fixed before merge.

Medium

internal/parser/types.go:367: Removing FindSourceFunc for OpenClaw/QClaw breaks on-disk session ID resolution in resolveRawSessionID, which still skips any file-based agent without FindSourceFunc. Unsynced OpenClaw/QClaw sessions passed to session usage or token-use will no longer be found or synced on demand.
- Fix: Add provider-backed lookup to resolveRawSessionID for provider-authoritative agents, or keep small FindSourceFunc adapters until those callers are migrated. Add on-disk resolution tests for OpenClaw and QClaw.

Panel: ci_default_security | Synthesis: codex, 19s | Members: codex_default (codex/default, done, 6m47s), codex_security (codex/security, done, 2m0s) | Total: 9m6s

roborev-ci · 2026-06-28T18:14:32Z

roborev: Combined Review (`ad8fb37`)

Reviewed provider migration: two medium regressions remain; no security findings were reported.

Medium

internal/parser/types.go:359: OpenClaw/QClaw now drop DiscoverFunc, but SSH remote sync still only emits remote transfer targets for file-based agents with DiscoverFunc != nil. Remote hosts with only OpenClaw/QClaw sessions will no longer transfer those directories, so new remote sessions for these agents silently stop syncing.
- Fix: Update the remote resolver to include provider-authoritative file-based agents via their default dirs/provider metadata, or keep temporary legacy discovery hooks until remote resolution is migrated; add a regression test that OpenClaw/QClaw appear in the resolve script.
internal/parser/types.go:367: Removing FindSourceFunc also removes OpenClaw/QClaw from resolveRawSessionID's on-disk probes, which still skip agents without that hook. A canonical or raw Claw ID that exists on disk but has not been synced yet now resolves as known=false, so session usage/token-use will not perform the intended on-demand sync.
- Fix: Add a provider-backed lookup path in resolveRawSessionID for provider-authoritative file-based agents, or retain compatibility FindSourceFunc adapters; cover unsynced OpenClaw/QClaw disk resolution in tests.

Panel: ci_default_security | Synthesis: codex, 10s | Members: codex_default (codex/default, done, 7m2s), codex_security (codex/security, done, 3m33s) | Total: 10m45s

roborev-ci · 2026-06-28T20:33:06Z

roborev: Combined Review (`f3f502a`)

The PR has one Medium issue to address before merging.

Medium

internal/parser/claw_provider.go:271: FindSource only resolves stored paths through currently configured roots. After this migration, SyncSingleSession can still prefer an existing DB file_path even when the OpenClaw/QClaw root was removed or changed, but processProviderFile then asks the provider to resolve that stored path and gets “provider source not found.” The legacy processors parsed that stored path directly, so explicit resyncs for existing sessions can regress.

Fix: Add a stored-path fallback for Claw sources that infers the implicit root from <root>/<agent>/sessions/<file>, validates the shape, and returns a SourceRef for existing stored paths outside configured roots.

Panel: ci_default_security | Synthesis: codex, 23s | Members: codex_default (codex/default, done, 7m23s), codex_security (codex/security, done, 1m56s) | Total: 9m42s

OpenClaw and QClaw share a Claw-style source layout where each agent directory owns a sessions folder and active JSONL files compete with archived JSONL variants for the same logical session. Moving them behind concrete provider facades keeps that active-over-archive and newest-archive policy explicit without broadening the generic JSONL source helpers around variable archive suffixes. The providers preserve colon-delimited agent/session lookup, selected-source change classification, symlinked agent directories, stale stored-path remapping, source fingerprinting, and existing parse normalization. fix(parser): promote claw archives on removal Claw providers choose a single source per logical session, so live-sync removal events need to account for source promotion. When an active file or newest archive disappears, another archive may become the selected source even though the changed path is no longer the source to parse. This keeps write events strict about the selected path, while remove and rename-style missing-path events can remap a valid stale Claw path to the newly selected source for the same raw session ID. test(parser): opt openclaw qclaw into provider shadow OpenClaw and QClaw now have concrete facade providers on this branch, so their migration modes should enter shadow comparison rather than staying legacy-only and additive. Earlier provider opt-ins remain inherited; later provider branches still own their own modes. Validation: go test -tags "fts5" ./internal/parser -run TestProviderMigrationModes -count=1; go test -tags "fts5" ./internal/parser -count=1; go vet ./...; git diff --check test(sync): compare claw shadow parity OpenClaw and QClaw are shadow-compared on this branch, so add source-level migration coverage that compares provider observation with their legacy parsers. The paired test follows the shared provider implementation and keeps the agent/session raw ID shape and planned data-version behavior visible during review. Validation: go test -tags "fts5" ./internal/parser ./internal/sync -run 'TestObserveProviderSourceMatchesClawLegacyParsers|Test(OpenClaw|QClaw)Provider|TestParse(OpenClaw|QClaw)' -count=1; go test -tags "fts5" ./internal/parser ./internal/sync -count=1; go fmt ./...; go vet ./...; git diff --check; ./custom-gcl run --config .golangci.nilaway.yml ./internal/parser/... ./internal/sync/... refactor(parser): fold claw providers into provider OpenClaw and QClaw should no longer keep exported discover/find/parse entrypoints beside the provider facade. Folding discovery, raw-ID lookup, archive selection, and parsing into the concrete providers makes this branch a real migration instead of another shim around the legacy path. The sync engine now relies on provider changed-path handling for this family, so the provider migration mode can become authoritative and the shadow-only comparison test is removed. Validation: go test -tags "fts5" ./internal/parser -run 'TestClawProvidersOwnLegacyEntrypoints|TestOpenClaw|TestQClaw|TestClawProvider|TestParseOpenClaw|TestParseQClaw|TestDiscoverOpenClaw|TestDiscoverQClaw|TestFindOpenClaw|TestFindQClaw' -count=1 -v; go test -tags "fts5" ./internal/sync -run 'TestEngine_ClassifyPathsQClaw|TestProviderMigration|TestObserveProvider|TestProviderProcess' -count=1 -v; go fmt ./...; go test -tags "fts5" ./internal/parser ./internal/sync ./cmd/agentsview -count=1; go vet ./...; git diff --check

Update openclaw and qclaw provider call sites to the exported source-set framework API.

clawSourceSet.Fingerprint built a SourceFingerprint without a Hash, so clawParseOutcome's guarded assignment left Session.File.Hash empty for both OpenClaw and QClaw. The legacy processOpenClaw/processQClaw paths always computed a full-file hash via ComputeFileHash and persisted file_hash, and the full-parse write overwrites file_hash unconditionally, so the migration cleared existing hashes to NULL on resync. Compute the content hash in Fingerprint via hashJSONLSourceFile.

mariusvniekerk changed the title ~~feat(parser): migrate claw providers~~ feat(parser): migrate OpenClaw + QClaw providers Jun 26, 2026

mariusvniekerk force-pushed the fam/qwen-family branch from 56db670 to e2596f2 Compare June 26, 2026 15:47

mariusvniekerk force-pushed the fam/claw-family branch from 5970de1 to 645790d Compare June 26, 2026 15:47

mariusvniekerk force-pushed the fam/qwen-family branch from e2596f2 to 9b703fc Compare June 26, 2026 18:55

mariusvniekerk force-pushed the fam/claw-family branch from 645790d to 08a36f5 Compare June 26, 2026 18:55

mariusvniekerk force-pushed the fam/qwen-family branch from 9b703fc to 8e3814e Compare June 26, 2026 19:07

mariusvniekerk force-pushed the fam/claw-family branch from 08a36f5 to 9cdc39d Compare June 26, 2026 19:07

mariusvniekerk force-pushed the fam/qwen-family branch from 8e3814e to 19a5630 Compare June 26, 2026 19:48

mariusvniekerk force-pushed the fam/claw-family branch from 9cdc39d to 78f2d96 Compare June 26, 2026 19:48

mariusvniekerk force-pushed the fam/qwen-family branch from 19a5630 to 830daba Compare June 26, 2026 19:54

mariusvniekerk force-pushed the fam/claw-family branch from 78f2d96 to 077f134 Compare June 26, 2026 19:54

mariusvniekerk force-pushed the fam/qwen-family branch from 830daba to f1b3f3a Compare June 26, 2026 20:32

mariusvniekerk force-pushed the fam/claw-family branch from 077f134 to 4f66903 Compare June 26, 2026 20:32

mariusvniekerk force-pushed the fam/claw-family branch from 4f66903 to 8eb63eb Compare June 27, 2026 12:50

mariusvniekerk force-pushed the fam/claw-family branch from 8eb63eb to def6e5d Compare June 27, 2026 23:02

mariusvniekerk force-pushed the fam/qwen-family branch from bf1bd46 to cf88fb4 Compare June 28, 2026 11:20

mariusvniekerk force-pushed the fam/claw-family branch from def6e5d to 48ae3e7 Compare June 28, 2026 11:20

mariusvniekerk force-pushed the fam/qwen-family branch from cf88fb4 to 9fa34de Compare June 28, 2026 14:08

mariusvniekerk force-pushed the fam/claw-family branch from 48ae3e7 to e15adbf Compare June 28, 2026 14:08

mariusvniekerk force-pushed the fam/qwen-family branch from 9fa34de to 522355d Compare June 28, 2026 15:10

mariusvniekerk force-pushed the fam/claw-family branch from e15adbf to edf4f0f Compare June 28, 2026 15:10

mariusvniekerk force-pushed the fam/qwen-family branch from 522355d to 71a061a Compare June 28, 2026 18:02

mariusvniekerk force-pushed the fam/claw-family branch from edf4f0f to ad8fb37 Compare June 28, 2026 18:02

mariusvniekerk force-pushed the fam/qwen-family branch from 71a061a to e58926d Compare June 28, 2026 20:20

mariusvniekerk force-pushed the fam/claw-family branch from ad8fb37 to f3f502a Compare June 28, 2026 20:20

mariusvniekerk force-pushed the fam/qwen-family branch from e58926d to 5362a46 Compare June 28, 2026 20:42

Base automatically changed from fam/qwen-family to main June 28, 2026 20:42

mariusvniekerk added 3 commits June 28, 2026 16:42

refactor(parser): adopt exported source-set API in claw providers

a8d793d

Update openclaw and qclaw provider call sites to the exported source-set framework API.

mariusvniekerk force-pushed the fam/claw-family branch from f3f502a to ebe3172 Compare June 28, 2026 20:42

mariusvniekerk merged commit a57f24f into main Jun 28, 2026
13 of 27 checks passed

mariusvniekerk deleted the fam/claw-family branch June 28, 2026 20:43

Uh oh!

Conversation

mariusvniekerk commented Jun 26, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

mariusvniekerk commented Jun 26, 2026

Uh oh!

roborev-ci Bot commented Jun 26, 2026

roborev: Combined Review (5970de1)

High

Medium

Uh oh!

roborev-ci Bot commented Jun 26, 2026

roborev: Combined Review (645790d)

Medium

Uh oh!

roborev-ci Bot commented Jun 26, 2026

roborev: Combined Review (08a36f5)

Medium

Uh oh!

roborev-ci Bot commented Jun 26, 2026

roborev: Combined Review (9cdc39d)

Medium

Uh oh!

roborev-ci Bot commented Jun 26, 2026

roborev: Combined Review (4f66903)

Medium

Uh oh!

roborev-ci Bot commented Jun 27, 2026

roborev: Combined Review (8eb63eb)

Medium

Uh oh!

roborev-ci Bot commented Jun 27, 2026

roborev: Combined Review (def6e5d)

Medium

Uh oh!

roborev-ci Bot commented Jun 28, 2026

roborev: Combined Review (48ae3e7)

Medium

Uh oh!

roborev-ci Bot commented Jun 28, 2026

roborev: Combined Review (e15adbf)

Uh oh!

roborev-ci Bot commented Jun 28, 2026

roborev: Combined Review (edf4f0f)

Medium

Uh oh!

roborev-ci Bot commented Jun 28, 2026

roborev: Combined Review (ad8fb37)

Medium

Uh oh!

roborev-ci Bot commented Jun 28, 2026

roborev: Combined Review (f3f502a)

Medium

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Milestone

Development

Uh oh!

1 participant

mariusvniekerk commented Jun 26, 2026 •

edited

Loading

roborev: Combined Review (`5970de1`)

roborev: Combined Review (`645790d`)

roborev: Combined Review (`08a36f5`)

roborev: Combined Review (`9cdc39d`)

roborev: Combined Review (`4f66903`)

roborev: Combined Review (`8eb63eb`)

roborev: Combined Review (`def6e5d`)

roborev: Combined Review (`48ae3e7`)

roborev: Combined Review (`e15adbf`)

roborev: Combined Review (`edf4f0f`)

roborev: Combined Review (`ad8fb37`)

roborev: Combined Review (`f3f502a`)