perf(#2351): batch path-existence checks via Git Trees API

.github/workflows/reusable-code.yml

@@ -178,4 +178,4 @@ jobs:
           run-url: ${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}
           status-repo: ${{ inputs.source_repo }}
           status-number: ${{ fromJSON(inputs.event_payload).issue.number }}
-          status-token: ${{ steps.app-token.outputs.token }}
+          mint-url: ${{ inputs.mint_url }}

.github/workflows/reusable-fix.yml

@@ -380,4 +380,4 @@ jobs:
           run-url: ${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}
           status-repo: ${{ inputs.source_repo }}
           status-number: ${{ steps.context.outputs.pr_number }}
-          status-token: ${{ steps.app-token.outputs.token }}
+          mint-url: ${{ inputs.mint_url }}

.github/workflows/reusable-retro.yml

@@ -153,4 +153,4 @@ jobs:
           run-url: ${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}
           status-repo: ${{ inputs.source_repo }}
           status-number: ${{ fromJSON(inputs.event_payload).pull_request.number || fromJSON(inputs.event_payload).issue.number }}
-          status-token: ${{ steps.app-token.outputs.token }}
+          mint-url: ${{ inputs.mint_url }}

.github/workflows/reusable-review.yml

@@ -169,4 +169,4 @@ jobs:
           run-url: ${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}
           status-repo: ${{ inputs.source_repo }}
           status-number: ${{ fromJSON(inputs.event_payload).pull_request.number || fromJSON(inputs.event_payload).issue.number }}
-          status-token: ${{ steps.app-token.outputs.token }}
+          mint-url: ${{ inputs.mint_url }}

.github/workflows/reusable-triage.yml

@@ -149,4 +149,4 @@ jobs:
           run-url: ${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}
           status-repo: ${{ inputs.source_repo }}
           status-number: ${{ fromJSON(inputs.event_payload).issue.number }}
-          status-token: ${{ steps.app-token.outputs.token }}
+          mint-url: ${{ inputs.mint_url }}

README.md

@@ -50,6 +50,7 @@ This is not a product spec. It's an evolving exploration of a hard problem space
   - [Vertex AI Inference Provisioning](docs/plans/vertex-inference-provisioning.md) — Provisioning and configuration for Vertex AI inference endpoints
   - [ADR-0045 Forge-Portable Harness Schema — Phase 1](docs/plans/adr-0045-forge-portable-harness-phase1.md) — Implementation plan for ADR-0045 forge-portable harness schema (Phase 1)
   - [ADR-0045 Forge-Portable Harness Schema — Phase 2](docs/plans/adr-0045-forge-portable-harness-phase2.md) — Implementation plan for ADR-0045 Phase 2: adopt new schema fields across install, scaffold, and lock flows
+  - [ADR-0045 Forge-Portable Harness Schema — Phase 3](docs/plans/adr-0045-forge-portable-harness-phase3.md) — Implementation plan for ADR-0045 Phase 3: deprecate config.yaml agents block, add Lint() diagnostics, migrate to harness-first discovery
   - [ADR-0046 Drift Scanner](docs/plans/2026-03-06-adr46-drift-scanner.md) — Implementation plan for ADR-0046 drift detection tool
 - **[docs/guides/](docs/guides/)** — Practical how-to documentation for administrators and developers (see [ADR 0023](docs/ADRs/0023-user-documentation-structure.md))
 - **[docs/ADRs/](docs/ADRs/)** — Architecture Decision Records for crystallizing specific decisions (see [ADR 0001](docs/ADRs/0001-use-adrs-for-decision-making.md))

action.yml

@@ -36,8 +36,16 @@ inputs:
   status-number:
     description: Issue/PR number for status comments (optional).
     default: ""
+  mint-url:
+    description: >-
+      Mint service URL for on-demand status comment tokens. When set, the
+      binary mints a fresh short-lived token before each status API call
+      instead of using a static status-token.
+    default: ""
   status-token:
-    description: Token for status comments (defaults to GH_TOKEN env var).
+    description: >-
+      DEPRECATED — use mint-url instead. Static GitHub token for status
+      comments. Ignored when mint-url is set.
     default: ""
 
 runs:
@@ -363,26 +371,31 @@ runs:
         STATUS_RUN_URL: ${{ inputs.run-url }}
         STATUS_REPO: ${{ inputs.status-repo }}
         STATUS_NUMBER: ${{ inputs.status-number }}
+        MINT_URL: ${{ inputs.mint-url }}
         STATUS_TOKEN: ${{ inputs.status-token }}
       run: |
         set -euo pipefail
+        if [[ -n "${STATUS_TOKEN}" ]]; then
+          echo "::add-mask::${STATUS_TOKEN}"
+        fi
         FULLSEND_DIR="${FULLSEND_DIR:-${GITHUB_WORKSPACE}}"
         TARGET_REPO="${TARGET_REPO:-${GITHUB_WORKSPACE}/target-repo}"
         mkdir -p "${GITHUB_WORKSPACE}/output"
         # SECURITY: Never expose --no-post-script as a workflow input.
         # Post-scripts enforce secret scanning, protected-path blocks,
         # and review-downgrade controls. Skipping them in CI bypasses
         # all post-push security gates.
-        if [[ -n "${STATUS_TOKEN}" ]]; then
-          echo "::add-mask::${STATUS_TOKEN}"
-        fi
         STATUS_FLAGS=()
         if [[ -n "${STATUS_REPO}" && -n "${STATUS_NUMBER}" ]]; then
           STATUS_FLAGS+=(--status-repo "${STATUS_REPO}" --status-number "${STATUS_NUMBER}")
           if [[ -n "${STATUS_RUN_URL}" ]]; then
             STATUS_FLAGS+=(--run-url "${STATUS_RUN_URL}")
           fi
+          if [[ -n "${MINT_URL}" ]]; then
+            STATUS_FLAGS+=(--mint-url "${MINT_URL}")
+          fi
           if [[ -n "${STATUS_TOKEN}" ]]; then
+            echo "::warning::status-token is deprecated; use mint-url instead"
             STATUS_FLAGS+=(--status-token "${STATUS_TOKEN}")
           fi
         fi
@@ -393,10 +406,12 @@ runs:
           "${STATUS_FLAGS[@]+"${STATUS_FLAGS[@]}"}"
 
     - name: Finalize orphaned status comment
-      if: always() && inputs.agent != '__install_only__' && inputs.status-repo != '' && inputs.status-number != ''
+      if: always() && inputs.agent != '__install_only__' && inputs.status-repo != '' && inputs.status-number != '' && (inputs.mint-url != '' || inputs.status-token != '')
       shell: bash
       env:
+        MINT_URL: ${{ inputs.mint-url }}
         STATUS_TOKEN: ${{ inputs.status-token }}
+        AGENT: ${{ inputs.agent }}
         STATUS_REPO: ${{ inputs.status-repo }}
         STATUS_NUMBER: ${{ inputs.status-number }}
         RUN_ID: ${{ github.run_id }}
@@ -405,17 +420,19 @@ runs:
         JOB_STATUS: ${{ job.status }}
       run: |
         set -euo pipefail
+        if [[ -n "${STATUS_TOKEN}" ]]; then
+          echo "::add-mask::${STATUS_TOKEN}"
+        fi
         # When the fullsend process is hard-killed (SIGKILL, OOM, segfault),
         # the deferred PostCompletion call never runs and the status comment
         # remains in "Started" state. This step runs unconditionally (if:
         # always()) to detect and finalize orphaned comments. See #2149.
-        TOKEN="${STATUS_TOKEN:-${GITHUB_TOKEN:-}}"
-        if [[ -z "${TOKEN}" ]]; then
-          echo "::warning::No token available for status comment reconciliation"
-          exit 0
+        RECONCILE_FLAGS=(--repo "${STATUS_REPO}" --number "${STATUS_NUMBER}" --run-id "${RUN_ID}")
+        if [[ -n "${MINT_URL}" ]]; then
+          RECONCILE_FLAGS+=(--mint-url "${MINT_URL}" --role "${AGENT}")
+        elif [[ -n "${STATUS_TOKEN}" ]]; then
+          RECONCILE_FLAGS+=(--token "${STATUS_TOKEN}")
         fi
-        echo "::add-mask::${TOKEN}"
-        RECONCILE_FLAGS=(--repo "${STATUS_REPO}" --number "${STATUS_NUMBER}" --run-id "${RUN_ID}" --token "${TOKEN}")
         if [[ -n "${RUN_URL}" ]]; then
           RECONCILE_FLAGS+=(--run-url "${RUN_URL}")
         fi

docs/ADRs/0045-forge-portable-harness-schema.md

@@ -142,8 +142,9 @@ agent definition `.md` file). `agent` describes *how* the agent behaves;
 `role` describes *what function* the agent serves in the pipeline; `slug`
 describes *who* the agent authenticates as. During Phase 1-2, `role` and
 `slug` are optional — `Validate()` does not require them. In Phase 3,
-`Validate()` emits warnings when `role` is missing. In Phase 4,
-`Validate()` requires `role`.
+`Validate()` continues to allow missing `role`, but `Lint()` emits
+warnings when `role` is missing. In Phase 4, `Validate()` requires
+`role`.
 
 `base` references another harness file whose fields serve as defaults for
 this harness. Any field set in the child overrides the corresponding base
@@ -516,11 +517,10 @@ func (h *Harness) ResolveForge(platform string) error { ... }
    Note: `role`/`slug` becoming required is independent of the `forge:`
    section — a harness that only targets one platform still needs `role`
    and `slug` but does not need `forge:`.
-   Implementation note: the current `Validate()` method returns hard errors
-   only — there is no warning/advisory path. Phase 3 will need a separate
-   `Lint()` method or log-level warnings to emit non-fatal diagnostics
-   without breaking existing callers that treat any `Validate()` error as
-   a hard stop.
+   Implementation note: `Validate()` returns hard errors only. Phase 3
+   adds a separate `Lint()` method that returns non-fatal `[]Diagnostic`
+   warnings without breaking existing callers that treat any `Validate()`
+   error as a hard stop.
 
 4. **Phase 4 (remove):** Require `role` in all harness files. Remove the
    `agents:` block from config.yaml entirely. Agent identity and

docs/agents/triage.md

@@ -40,14 +40,45 @@ outcome and the post-script applies the corresponding label.
 | `ready-to-code` | The issue is fully specified and low-risk (bug, documentation, performance). Triggers the [code agent](code.md). |
 | `triaged` | The issue is fully specified but is a feature or other category that requires human prioritization before coding. |
 | `duplicate` | The issue duplicates an existing one. The agent identified the original and the post-script closes the issue. |
-| `blocked` | The issue depends on another issue or external condition. The agent identified the blocker. |
+| `blocked` | The issue depends on prerequisites — existing issues/PRs or newly created upstream issues. The agent identified or created the blockers. |
 | `question` | The issue is a support request or question, not an actionable bug or feature. The agent attempted to answer it. |
 
 The `issue-labels` skill may also apply contextual labels (e.g., `area/api`,
 `kind/bug`) but these are informational — they do not control agent behavior.
 
 ## Configuration and extension
 
+### Cross-repo issue creation
+
+The triage agent can create prerequisite issues in other repositories when it
+identifies upstream dependencies that don't have tracking issues yet. This is
+controlled by the `create_issues` section in `config.yaml`:
+
+```yaml
+create_issues:
+  allow_targets:
+    orgs:
+      - my-org
+    repos:
+      - upstream-org/specific-repo
+```
+
+**Defaults:** At install time, fullsend populates this with your org (in org mode)
+or your repo (in per-repo mode), plus `fullsend-ai/fullsend` as an upstream target.
+
+**When to expand the allowlist:** If your project depends on libraries or services
+in other GitHub orgs and you want the triage agent to automatically file
+prerequisite issues there, add those orgs or repos to `allow_targets`.
+
+**When to restrict the allowlist:** If you don't want agents creating issues
+outside your org, remove entries. If `allow_targets` is empty, automatic
+prerequisite creation is disabled entirely — the agent will still identify
+the dependency and include a draft issue body in its comment for a human to
+file manually.
+
+The source repo (where triage is running) is always implicitly allowed
+regardless of the allowlist.
+
 ### Skill: `issue-labels`
 
 The triage agent includes a built-in `issue-labels` skill that discovers your

docs/architecture.md

@@ -236,7 +236,7 @@ ADR 0002: [Building block 3](ADRs/0002-initial-fullsend-design.md#3-label-state-
 
 ### 4. triage agent runtime
 
-Runs triage from issue `title`/`body` + GitHub-native attachments only; each run starts with **`duplicate`** and other reset labels cleared; duplicate detection, blocking dependency detection (cross-repo), readiness, reproducibility, test handoff; can close as duplicate again if still a match, or label **`blocked`** when progress depends on another open issue or PR.
+Runs triage from issue `title`/`body` + GitHub-native attachments only; each run starts with **`duplicate`** and other reset labels cleared; duplicate detection, prerequisite detection (cross-repo), readiness, reproducibility, test handoff; can close as duplicate again if still a match, label **`blocked`** when progress depends on another open issue or PR, or create upstream prerequisite issues when no tracking issue exists (controlled by `create_issues.allow_targets` config).
 ADR 0002: [Building block 4](ADRs/0002-initial-fullsend-design.md#4-triage-agent-runtime).
 
 ### 5. Duplicate / similarity search

docs/guides/dev/cli-internals.md

@@ -58,7 +58,7 @@ fullsend
 │   ├── --run-url <url>                      #   CI/CD run URL for status comments
 │   ├── --status-repo <owner/repo>           #   Repository for status comments
 │   ├── --status-number <int>                #   Issue/PR number for status comments
-│   └── --status-token <token>               #   Token for status comments (default: GH_TOKEN)
+│   └── --mint-url <url>                     #   Mint service URL for on-demand status tokens
 ├── fetch-skill      <url>                    # Fetch a skill at runtime (in-sandbox)
 ├── scan                                     # Run security scanner on input/output
 │   ├── input                                # Scan event payload for prompt injection
@@ -74,7 +74,8 @@ fullsend
     ├── --run-url <url>                      #   Workflow run URL (optional)
     ├── --sha <string>                       #   Commit SHA (optional)
     ├── --reason <string>                    #   Termination reason: terminated or cancelled (default: terminated)
-    └── --token <token>                      #   GitHub token (default: $GITHUB_TOKEN)
+    ├── --mint-url <url>                     #   Mint service URL for on-demand token (default: $FULLSEND_MINT_URL)
+    └── --role <string>                      #   Agent role for minting (required with --mint-url)
 ```
 
 ### Command Decomposition

docs/guides/user/bugfix-workflow.md

@@ -102,7 +102,7 @@ Every push to a PR in the review stage triggers a new review round. This means `
 The triage agent:
 
 1. **Checks for duplicates.** Searches existing issues by title, body, and metadata. If it finds a match with high confidence, it labels `duplicate`, posts a comment linking the canonical issue, and closes this one.
-2. **Checks for blocking dependencies.** Searches for open issues or PRs (in this repo or upstream) that must be resolved before work can start. If a blocker is found, it labels `blocked` and posts a comment linking to the blocking issue or PR. On re-triage, it checks whether existing blockers have been resolved.
+2. **Checks for blocking dependencies.** Searches for open issues or PRs (in this repo or upstream) that must be resolved before work can start. If a prerequisite is found, it labels `blocked` and posts a comment linking to it. When no upstream tracking issue exists, the triage agent can also create one in the upstream repo (controlled by `create_issues.allow_targets` in config). On re-triage, it checks whether existing prerequisites have been resolved.
 3. **Checks information sufficiency.** If the issue body is missing steps to reproduce, expected behavior, or other critical details, it labels `needs-info` and posts a comment explaining what's missing.
 4. **Produces a test artifact.** When possible, writes a failing test case aligned with the repo's test framework.
 5. **Hands off.** Labels `ready-to-code` with a summary comment.

docs/guides/user/running-agents-locally.md

@@ -235,7 +235,7 @@ target issue/PR. These flags mirror what the CI workflows pass automatically:
 | `--run-url` | URL of the CI/CD run shown in the status comment |
 | `--status-repo` | Repository (`owner/repo`) to post status comments on |
 | `--status-number` | Issue or PR number for status comments |
-| `--status-token` | Token for posting comments (defaults to `GH_TOKEN`) |
+| `--mint-url` | Mint service URL for on-demand status comment tokens (default: `$FULLSEND_MINT_URL`) |
 
 Example: