Skip to content

docs(adr): define repository-less agent execution - #8702

Closed
csoceanu wants to merge 5 commits into
fullsend-ai:mainfrom
csoceanu:codex/aisdlc-146-adr
Closed

csoceanu wants to merge 5 commits into
fullsend-ai:mainfrom
csoceanu:codex/aisdlc-146-adr

Conversation

@csoceanu

@csoceanu csoceanu commented Oct 22, 2026 •

Copy link
Copy Markdown
Contributor

Summary

  • Add ADR 0150 defining repository-less agent execution.
  • Separate the sandbox workspace, execution directory, and optional target repository.
  • Document harness capability, forge scope, and current behavior with SHA-pinned references.
  • Annotate ADRs 0098 and 0112 where this decision narrows existing semantics.
  • Update the architecture and agent-infrastructure documents.
  • This PR contains documentation and design only; runtime implementation is intentionally deferred.

Related tickets

  • AISDLC-455 (parent: AISDLC-146)

Validation

  • make lint
  • git diff --check
  • ADR status, numbering, frontmatter, and documentation-link checks

@qodo-code-review

qodo-code-review Bot commented Oct 22, 2026 •

Copy link
Copy Markdown

Code Review by Qodo

🐞 Bugs (0) 📘 Rule violations (0) 📎 Requirement gaps (0)

Grey Divider

Great, no issues found!

Qodo reviewed your code and found no material issues that require review

Grey Divider

Tip of the day
💡 Did you know, you can ask Qodo to dismiss a finding you disagree with, with your reason on record

More tips ↗ | Customize Qodo ↗ | Qodo docs ↗

Grey Divider

Previous reviews

Review updated until commit a547fae ⏭️ Skipped

Results up to commit 26500f9 ⏭️ Skipped


No changes from previous review

Grey Divider

Qodo Logo

@Benkapner

Copy link
Copy Markdown
Member

/fs-fix-stop

@qodo-code-review

qodo-code-review Bot commented Oct 22, 2026 •

Copy link
Copy Markdown

PR Summary by Qodo

Define the repository-less agent execution contract

📝 Documentation 🕐 20-40 Minutes

Grey Divider

AI Description

• Define explicit repository-less execution while preserving existing repository-targeted behavior.
• Separate the sandbox workspace, execution directory, and optional target repository to prevent
 implicit repository authority.
• Document harness opt-in, forge boundaries, attribution, and deferred implementation requirements.
Diagram

graph TD
  CLI["Run invocation"] --> Config["Configuration root"] --> Harness["Resolved harness"] --> Mode{"Target supplied?"}
  Mode --> Workspace["Sandbox workspace"] --> Exec["Execution directory"]
  Mode -->|"explicit target"| Repo["Target repository"] --> Exec
Loading
High-Level Assessment

The following are alternative approaches to this PR:

1. Separate repository-less runner
  • ➕ Could isolate repository-less orchestration from existing target-specific behavior.
  • ➖ Would duplicate configuration loading and harness resolution, risking divergent policy and attribution behavior.

Recommendation: Keep the ADR's shared-runner contract. Explicit mode selection and a separately passed execution directory preserve existing targeted behavior without giving scratch space repository identity; the separate-runner alternative adds duplication without resolving that boundary.

Files changed (6) +238 / -2

Documentation (6) +238 / -2
0098-entity-first-harness-evaluation.mdClarify attribution without a target repository +6/-0

Clarify attribution without a target repository

• Annotates the existing entity-first decision so repository-less runs retain invocation, policy, harness, and entity attribution without a sentinel repository.

docs/ADRs/0098-entity-first-harness-evaluation.md

0112-overlays-may-set-any-harness-field.mdExclude repository capability from overlays +5/-0

Exclude repository capability from overlays

• Records that the planned target_repo_required field is composed at the harness top level and cannot be changed through overlays or legacy forge blocks.

docs/ADRs/0112-overlays-may-set-any-harness-field.md

0150-repository-less-agent-execution.mdEstablish the repository-less execution ADR +198/-0

Establish the repository-less execution ADR

• Defines explicit CLI modes, distinct workspace and repository identities, harness opt-in, and boundaries for forge selection, credentials, status, and attribution. Documents current limitations with pinned source references and sets acceptance criteria for later implementation; this PR does not implement the runtime behavior.

docs/ADRs/0150-repository-less-agent-execution.md

architecture.mdAdd repository-less execution to the architecture +6/-0

Add repository-less execution to the architecture

• Describes the accepted target-state separation of sandbox workspace, execution directory, and optional repository, with an ADR link.

docs/architecture.md

harness-fields.mdDocument the planned harness capability field +3/-1

Document the planned harness capability field

• Classifies target_repo_required as a planned top-level boolean and explains its default, base-composition precedence, and exclusion from overlays and forge blocks.

docs/contributing/harness-fields.md

agent-infrastructure.mdLink the infrastructure question to the new decision +20/-1

Link the infrastructure question to the new decision

• Separates the decided repository-less workspace contract from still-open external resource access, while retaining links to existing credential and tool-access decisions.

docs/problems/agent-infrastructure.md

@github-actions github-actions Bot added the fullsend-no-fix Skip bot-triggered fix agent runs label Oct 22, 2026
@github-actions

Copy link
Copy Markdown

Fix agent disabled for this PR. Remove the fullsend-no-fix label or use /fs-fix to re-engage.

@fullsend-ai-review

fullsend-ai-review Bot commented Oct 22, 2026 •

Copy link
Copy Markdown

🤖 Review · ⚠️ Cancelled · Started 11:26 AM UTC · Ended 11:29 AM UTC

Commit: 26500f9 · View workflow run →

This review was superseded by a newer commit. Check the current pull request HEAD's review status before merging.

@github-actions

github-actions Bot commented Oct 22, 2026 •

Copy link
Copy Markdown

Site preview

Preview: https://a23de6a7-site.fullsend-ai.workers.dev

Commit: fb05052f37bebee7cfc3047f5efa79c28feebd07

@fullsend-ai-review

Copy link
Copy Markdown

🤖 Finished Review · ⚠️ Cancelled · Started 11:26 AM UTC · Completed 11:29 AM UTC

Commit: 26500f9 · View workflow run →

Runtime: pi · Model: sonnet → claude-sonnet-5-5 · Effort: high · Cost: $0.46

This review was superseded by a newer commit. Check the current pull request HEAD's review status before merging.

@Benkapner

Copy link
Copy Markdown
Member

/fs-review

@fullsend-ai-review

Copy link
Copy Markdown

🤖 Review · ⚠️ Cancelled · Ended 11:31 AM UTC

Commit: b0c46eb · View workflow run →

Automated review did not complete for this commit. Check the latest review status before merging. If no review is queued or running, comment /fs-review to retry.

@fullsend-ai-review

fullsend-ai-review Bot commented Oct 22, 2026 •

Copy link
Copy Markdown

🤖 Finished Review · ⚠️ Cancelled · Started 11:33 AM UTC · Completed 11:36 AM UTC

Commit: b0c46eb · View workflow run →

Runtime: pi · Model: sonnet → claude-sonnet-5-5 · Effort: high · Cost: $0.74

This review was superseded by a newer commit. Check the current pull request HEAD's review status before merging.

@fullsend-ai-review

fullsend-ai-review Bot commented Oct 22, 2026 •

Copy link
Copy Markdown

🤖 Finished Review · ✅ Success · Started 11:38 AM UTC · Completed 11:54 AM UTC

Commit: 14dc55c · View workflow run →

Runtime: pi · Model: sonnet → claude-sonnet-5-5 · Effort: high · Cost: $2.75

@rh-hemartin

rh-hemartin commented Oct 22, 2026 •

Copy link
Copy Markdown
Member

First I would like you to simplify some paragraphs, they are very dense and with the typical language of AI. Make them simpler please.

Second... can't we have the same results with an empty directory and an agent that does not expect to go back to a repository? In other words, I would like some use cases described that justify this. I could create a weather-report agent, pass an empty directory as --target-repo and make it write a weather report in a markdown file that is not committed. Maybe it could post that text into an issue without need of repository. Does renaming --target-repo to --working-directory helps? The examples can be just comments on this PR, I don't need them on the ADR.

@codecov

codecov Bot commented Oct 22, 2026

Copy link
Copy Markdown

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!

@csoceanu
csoceanu marked this pull request as draft October 22, 2026 11:51
@fullsend-ai-review fullsend-ai-review Bot added the risk/moderate PR risk: moderate label Oct 22, 2026
@fullsend-ai-review

fullsend-ai-review Bot commented Oct 22, 2026 •

Copy link
Copy Markdown

<!-- fullsend:risk-assessment -->
Risk Assessment: moderate (2/5)

Details

Documentation-only addition of ADR 0150 by an established author with no protected paths or code changes, elevated to moderate risk by high churn and file coupling on central architecture documentation.

Previous run

<!-- fullsend:risk-assessment -->
Risk Assessment: moderate (2/5)

Details

Documentation-only addition of ADR 0150 by an established author with no protected paths or code changes, elevated to moderate risk by high churn and file coupling on central architecture documentation.

Previous run (2)

<!-- fullsend:risk-assessment -->
Risk Assessment: moderate (2/5)

Details

Documentation-only addition of ADR 0150 by an established author with no protected paths or code changes, elevated to moderate risk by high churn and file coupling on central architecture documentation.

Previous run (3)

<!-- fullsend:risk-assessment -->
Risk Assessment: moderate (2/5)

Details

Documentation-only addition of ADR 0150 by an established author with no protected paths or code changes, elevated to moderate risk by high churn and file coupling on central architecture documentation.

Previous run (4)

<!-- fullsend:risk-assessment -->
Risk Assessment: moderate (2/5)

Details

Documentation-only addition of ADR 0150 by an established author with no protected paths or code changes, elevated to moderate risk by high churn and file coupling on central architecture documentation.

Previous run (5)

<!-- fullsend:risk-assessment -->
Risk Assessment: moderate (2/5)

Details

Documentation-only addition of ADR 0150 by an established author with no protected paths or code changes, elevated to moderate risk by high churn and file coupling on central architecture documentation.

Previous run (6)

<!-- fullsend:risk-assessment -->
Risk Assessment: moderate (2/5)

Details

Documentation-only addition of ADR 0150 by an established author with no protected paths or code changes, elevated to moderate risk by high churn and file coupling on central architecture documentation.

@fullsend-ai-review

fullsend-ai-review Bot commented Oct 22, 2026 •

Copy link
Copy Markdown

Looks good to me

Earlier findings

  • docs/ADRs/0150-repository-less-agent-execution.md:1, missing-authorization, reclassified (remains info): the PR still has no linked GitHub issue, only the external Jira reference AISDLC-455; advisory only.

No findings at or above low severity; 1 lower-severity finding(s) were not posted.

Previous run

<!-- fullsend:review-findings-v2:eyJ2ZXJzaW9uIjoyLCJhY3Rpb24iOiJhcHByb3ZlIiwiZmluZGluZ3MiOlt7InNldmVyaXR5IjoiaW5mbyIsImNhdGVnb3J5IjoibWlzc2luZy1hdXRob3JpemF0aW9uIiwiZmlsZSI6ImRvY3MvQURScy8wMTUwLXJlcG9zaXRvcnktbGVzcy1hZ2VudC1leGVjdXRpb24ubWQiLCJpZCI6ImZfZGI4NmMyN2UxMmY2Mzc3OCIsImxpbmUiOjEsImFjdGlvbmFibGUiOmZhbHNlLCJkaW1lbnNpb24iOiJpbnRlbnQtY29oZXJlbmNlIn0seyJzZXZlcml0eSI6ImxvdyIsImNhdGVnb3J5IjoiZG9jLXN0eWxlIiwiZmlsZSI6ImRvY3MvQURScy8wMDk4LWVudGl0eS1maXJzdC1oYXJuZXNzLWV2YWx1YXRpb24ubWQiLCJpZCI6ImZfM2UxMDZjMTA1YThjNGM0OSIsImxpbmUiOjI0LCJhY3Rpb25hYmxlIjp0cnVlfSx7InNldmVyaXR5IjoiaW5mbyIsImNhdGVnb3J5IjoiZG9jLXN0eWxlIiwiZmlsZSI6ImRvY3MvQURScy8wMTEyLW92ZXJsYXlzLW1heS1zZXQtYW55LWhhcm5lc3MtZmllbGQubWQiLCJpZCI6ImZfNGMxZjU0OTQzYjk0MmRkMyIsImxpbmUiOjI0LCJhY3Rpb25hYmxlIjpmYWxzZX0seyJzZXZlcml0eSI6ImluZm8iLCJjYXRlZ29yeSI6InBhdHRlcm4taW5jb25zaXN0ZW5jeSIsImZpbGUiOiJkb2NzL2FyY2hpdGVjdHVyZS5tZCIsImlkIjoiZl8yNmYxMzc2ZWIzYzYxNmM2IiwibGluZSI6NTMsImFjdGlvbmFibGUiOmZhbHNlfSx7InNldmVyaXR5IjoiaW5mbyIsImNhdGVnb3J5IjoiZG9jLXN0eWxlIiwiZmlsZSI6ImRvY3MvQURScy8wMTUwLXJlcG9zaXRvcnktbGVzcy1hZ2VudC1leGVjdXRpb24ubWQiLCJpZCI6ImZfNTI2MGZlNmM2YzEzYmZjMCIsImxpbmUiOjExNCwiYWN0aW9uYWJsZSI6ZmFsc2V9LHsic2V2ZXJpdHkiOiJpbmZvIiwiY2F0ZWdvcnkiOiJwYXR0ZXJuLWluY29uc2lzdGVuY3kiLCJmaWxlIjoiZG9jcy9jb250cmlidXRpbmcvaGFybmVzcy1maWVsZHMubWQiLCJpZCI6ImZfYmQyN2Y3NjlmN2ZmNjM1MSIsImxpbmUiOjY3LCJhY3Rpb25hYmxlIjpmYWxzZX1dLCJkaXNwb3NpdGlvbnMiOlt7ImlkIjoiZl9kYjg2YzI3ZTEyZjYzNzc4Iiwic3RhdHVzIjoicmVjbGFzc2lmaWVkIn0seyJpZCI6ImZfM2UxMDZjMTA1YThjNGM0OSIsInN0YXR1cyI6InJlc29sdmVkX2J5X2NoYW5nZSJ9LHsiaWQiOiJmXzRjMWY1NDk0M2I5NDJkZDMiLCJzdGF0dXMiOiJyZXNvbHZlZF9ieV9jaGFuZ2UifSx7ImlkIjoiZl8yNmYxMzc2ZWIzYzYxNmM2Iiwic3RhdHVzIjoicmVzb2x2ZWRfYnlfY2hhbmdlIn0seyJpZCI6ImZfNTI2MGZlNmM2YzEzYmZjMCIsInN0YXR1cyI6InJlc29sdmVkX2J5X2NoYW5nZSJ9LHsiaWQiOiJmX2JkMjdmNzY5ZjdmZjYzNTEiLCJzdGF0dXMiOiJyZXNvbHZlZF9ieV9jaGFuZ2UifV19 -->

Looks good to me


No findings at or above low severity; 1 lower-severity finding(s) were not posted.

Previous run (2)

<!-- fullsend:review-findings-v2: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 -->

Looks good to me

Previous run (3)

<!-- fullsend:review-findings-v2:eyJ2ZXJzaW9uIjoyLCJhY3Rpb24iOiJhcHByb3ZlIiwiZmluZGluZ3MiOlt7InNldmVyaXR5IjoiaW5mbyIsImNhdGVnb3J5IjoiZG9jLXN0eWxlIiwiZmlsZSI6ImRvY3MvQURScy8wMTEyLW92ZXJsYXlzLW1heS1zZXQtYW55LWhhcm5lc3MtZmllbGQubWQiLCJpZCI6ImZfNGMxZjU0OTQzYjk0MmRkMyIsImxpbmUiOjI0LCJhY3Rpb25hYmxlIjpmYWxzZX0seyJzZXZlcml0eSI6ImluZm8iLCJjYXRlZ29yeSI6InBhdHRlcm4taW5jb25zaXN0ZW5jeSIsImZpbGUiOiJkb2NzL2FyY2hpdGVjdHVyZS5tZCIsImlkIjoiZl8yNmYxMzc2ZWIzYzYxNmM2IiwibGluZSI6NTMsImFjdGlvbmFibGUiOmZhbHNlfSx7InNldmVyaXR5IjoiaW5mbyIsImNhdGVnb3J5IjoiZG9jLXN0eWxlIiwiZmlsZSI6ImRvY3MvQURScy8wMTUwLXJlcG9zaXRvcnktbGVzcy1hZ2VudC1leGVjdXRpb24ubWQiLCJpZCI6ImZfNTI2MGZlNmM2YzEzYmZjMCIsImxpbmUiOjExNCwiYWN0aW9uYWJsZSI6ZmFsc2V9LHsic2V2ZXJpdHkiOiJpbmZvIiwiY2F0ZWdvcnkiOiJwYXR0ZXJuLWluY29uc2lzdGVuY3kiLCJmaWxlIjoiZG9jcy9jb250cmlidXRpbmcvaGFybmVzcy1maWVsZHMubWQiLCJpZCI6ImZfYmQyN2Y3NjlmN2ZmNjM1MSIsImxpbmUiOjY3LCJhY3Rpb25hYmxlIjpmYWxzZX0seyJzZXZlcml0eSI6ImluZm8iLCJjYXRlZ29yeSI6Im1pc3NpbmctYXV0aG9yaXphdGlvbiIsImZpbGUiOiJkb2NzL0FEUnMvMDE1MC1yZXBvc2l0b3J5LWxlc3MtYWdlbnQtZXhlY3V0aW9uLm1kIiwiaWQiOiJmX2RiODZjMjdlMTJmNjM3NzgiLCJsaW5lIjoxLCJhY3Rpb25hYmxlIjpmYWxzZX0seyJzZXZlcml0eSI6ImxvdyIsImNhdGVnb3J5IjoiZG9jLXN0eWxlIiwiZmlsZSI6ImRvY3MvQURScy8wMDk4LWVudGl0eS1maXJzdC1oYXJuZXNzLWV2YWx1YXRpb24ubWQiLCJpZCI6ImZfM2UxMDZjMTA1YThjNGM0OSIsImxpbmUiOjI0LCJhY3Rpb25hYmxlIjp0cnVlfV0sImRpc3Bvc2l0aW9ucyI6W3siaWQiOiJmXzNlMTA2YzEwNWE4YzRjNDkiLCJzdGF0dXMiOiJyZXNvbHZlZF9ieV9jaGFuZ2UifSx7ImlkIjoiZl80YzFmNTQ5NDNiOTQyZGQzIiwic3RhdHVzIjoicmVjbGFzc2lmaWVkIn0seyJpZCI6ImZfMjZmMTM3NmViM2M2MTZjNiIsInN0YXR1cyI6InJlY2xhc3NpZmllZCJ9LHsiaWQiOiJmXzUyNjBmZTZjNmMxM2JmYzAiLCJzdGF0dXMiOiJyZWNsYXNzaWZpZWQifSx7ImlkIjoiZl9iZDI3Zjc2OWY3ZmY2MzUxIiwic3RhdHVzIjoicmVjbGFzc2lmaWVkIn0seyJpZCI6ImZfZGI4NmMyN2UxMmY2Mzc3OCIsInN0YXR1cyI6InJlY2xhc3NpZmllZCJ9XX0= -->

Looks good to me

Earlier findings

  • docs/ADRs/0098-entity-first-harness-evaluation.md:24, doc-style — resolved: the ADR 0150 annotation paragraph was rewrapped in the latest commit.
Previous run (4)

<!-- fullsend:review-findings-v2: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 -->

Review

Findings

Low

  • [doc-style] docs/ADRs/0098-entity-first-harness-evaluation.md:24 — The added ADR annotation wraps irregularly: line 24 (for repository-less execution: the target) ends near column 41 and splits "target repository" across lines, while the next line runs to ~79 columns. docs/contributing/documentation.md asks for prose wrapped close to 80 columns.
    Remediation: Rewrap the annotation paragraph close to 80 columns (e.g. break after "the target repository may be absent, while").

Earlier findings

  • docs/ADRs/0112-overlays-may-set-any-harness-field.md:24, doc-style — reclassified to info: the early wrap after "adds" is a borderline nit under the ~80 column guide.

Next steps:

  • /fs-fix — agent addresses review findings automatically
  • /fs-fix <your instruction> — agent fixes with your specific guidance
  • Push commits directly — review re-runs automatically on push
  • /fs-fix-stop — disable automatic fix runs for this PR
Previous run (5)

<!-- fullsend:review-findings-v2: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 -->

Review

Findings

Low

  • [doc-style] docs/ADRs/0098-entity-first-harness-evaluation.md:24 — Uneven line wrapping: line 24 breaks at 41 columns mid-phrase ("for repository-less execution: the target") while the next line runs to 79, contrary to the near-80-column rule in docs/contributing/documentation.md.
    Remediation: Rewrap lines 23-27 at word boundaries close to 80 columns.
  • [doc-style] docs/ADRs/0112-overlays-may-set-any-harness-field.md:24 — Premature line break: line 23 ends after "adds" at 56 columns although `target_repo_required` would still fit within 80.
    Remediation: Rewrap the annotation paragraph so `target_repo_required` continues on line 23.

Earlier findings

  • PR-level / docs/ADRs/0150-repository-less-agent-execution.md:1, missing-authorization — reclassified to info: docs/contributing/adrs.md says a new ADR is approved at merge and the PR cites its tracker ticket, so a linked GitHub issue is not required.
  • docs/architecture.md:53, pattern-inconsistency — reclassified to info: other planned entries in the same file use future tense, and splitting an ADR link across lines is common in the repo.
  • docs/ADRs/0150-repository-less-agent-execution.md:113, doc-style — reclassified to info: the cited line is blank and the nearby bullets wrap at word boundaries under 80 columns.
  • docs/contributing/harness-fields.md:67, pattern-inconsistency — reclassified to info: the row does state why the field is top-level only; only a duplicated "Planned" remains.

Next steps:

  • /fs-fix — agent addresses review findings automatically
  • /fs-fix <your instruction> — agent fixes with your specific guidance
  • Push commits directly — review re-runs automatically on push
  • /fs-fix-stop — disable automatic fix runs for this PR
Previous run (6)

<!-- fullsend:review-findings-v2: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 -->

Review

Findings

Medium

Low

  • [doc-style] docs/ADRs/0098-entity-first-harness-evaluation.md:23 — The new annotation names "ADR 0150" as plain text, while the next paragraph links its ADR reference.
    Remediation: Change to [ADR 0150](0150-repository-less-agent-execution.md).
  • [doc-style] docs/ADRs/0112-overlays-may-set-any-harness-field.md:24 — The new annotation names "ADR 0150" as plain text; the first mention of every other ADR in this file is hyperlinked.
    Remediation: Change to [ADR 0150](0150-repository-less-agent-execution.md).
  • [pattern-inconsistency] docs/architecture.md:53 — Other **Decided:** entries lead with the topic and use a (target state) qualifier for unimplemented decisions (lines 396, 454, 464). The new bullet starts with a bold **Not yet implemented:** prefix used nowhere else in the file.
    Remediation: Reword as - Repository-less execution (target state): will use an explicit --no-target-repo mode... ([ADR 0150](ADRs/0150-repository-less-agent-execution.md)).
  • [doc-style] docs/ADRs/0150-repository-less-agent-execution.md:113 — Lines 113 ("particular,") and 136 ("repository-less") each hold a single orphaned word mid-paragraph, inconsistent with the ~80-column wrapping used elsewhere.
    Remediation: Rewrap lines 112-114 and 135-137 so lines fill to about 80 columns.
  • [pattern-inconsistency] docs/contributing/harness-fields.md:67 — Sibling planned rows cite their source ADRs (schema_version (ADR-0127), preflight_check (ADR-0128, ADR-0129)) and end with **Planned, not yet implemented** without a period inside the bold. The new target_repo_required row cites no ADR and puts a period inside the bold.
    Remediation: Add (ADR-0150) and drop the period inside the bold to match the siblings.

Labels: PR changes documentation only (ADRs, architecture and contributor docs)


Next steps:

  • /fs-fix — agent addresses review findings automatically
  • /fs-fix <your instruction> — agent fixes with your specific guidance
  • Push commits directly — review re-runs automatically on push
  • /fs-fix-stop — disable automatic fix runs for this PR

@rh-hemartin

rh-hemartin commented Oct 30, 2026 •

Copy link
Copy Markdown
Member

In the case of rfe-creator the fact that is passing its own checkout is because a restriction in the harness about skills. The harness of rfe-creator lives on /.fullsend/rfe-creator/ but the skills live on /.claude/skills, so the harness can't reference them as it is exiting the .fullsend directory (../../.claude/skills/<skill-name>), where it is "jailed".

This solves nothing, just adding information 🤣

@rh-hemartin

Copy link
Copy Markdown
Member

I can see both ways, so I'm fine with either. A compromise would be to say it is experimental and that could be removed at any time, just to justify a later removal if we feel this is not really used. Also this is not blocking the use case, they need just to provide an empty directory, so this is not that pressing.

Clarify source and target entity boundaries, preserve Git-managed configuration, and mark repository-less execution experimental.

Co-authored-by: Codex <noreply@openai.com>
Signed-off-by: csoceanu <csoceanu@redhat.com>
@fullsend-ai-review

fullsend-ai-review Bot commented Nov 2, 2026 •

Copy link
Copy Markdown

🤖 Finished Review · ✅ Success · Started 8:25 AM UTC · Completed 8:37 AM UTC

Commit: b321c9d · View workflow run →

Runtime: pi · Model: sonnet → claude-sonnet-5-5 · Effort: high · Cost: $1.92

@fullsend-ai-review fullsend-ai-review Bot removed the ready-for-merge All reviewers approved — ready to merge label Nov 2, 2026
fullsend-ai-review[bot]

This comment was marked as outdated.

@fullsend-ai-review
fullsend-ai-review Bot dismissed their stale review November 2, 2026 08:37

Superseded by updated review

@fullsend-ai-review fullsend-ai-review Bot added the ready-for-merge All reviewers approved — ready to merge label Nov 2, 2026
@csoceanu

csoceanu commented Nov 2, 2026

Copy link
Copy Markdown
Contributor Author

Thanks a lot for your feedback! I pushed b321c9d with the latest upstream merge and clarified ADR 0150 based on the discussion.
The ADR now makes clear that this is not about removing Git or configuration repositories. It separates the source/invocation entity from the optional target repository, and marks the capability as experimental.
Hector’s note about the rfe-creator skills restriction is a separate implementation concern and is not claimed to be solved by this ADR.
@rh-hemartin @ifireball, could you take another look?

which is execution-host metadata, not proof that the host checkout is the
target. Treating that checkout as a target would conflate configuration,
execution, code, status, and credential scope. Some Jira, tenant, and
service-level tasks have configuration and input but no target code repository.

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

So ultimately we want to get read of the mandatory --target-repo parameter? I hink we need an explanation somewhere about why we cant leave it as-is and just ignore what its doing.

@csoceanu csoceanu Nov 3, 2026 •

Copy link
Copy Markdown
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

thanks, in fb05052 I added an explicit rejection for treating --target-repo as only a working directory. An empty directory may let a simple agent start, but it still goes through the target-repository path. The ADR keeps the explicit --no-target-repo mode from AISDLC-146 so execution and target identity stay separate.

a target code repository while preserving existing repository-targeted
behavior. This is an experimental capability: it records a contract for a
follow-up implementation and does not make every built-in agent
repository-independent.

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

So while I understand the scope better now I still wonder, do we want agent without a target at all or is the case we're looking at is one where the agent simple has a target that isn't a repo.

For the one case I can think of where the agent isn't using the target repo, its basically "triage" which write back to the issue it was invoked from, so that does seem like a target-free agent, or perhaps we should consider that to be an agent where bot the "source" and the "target" happen to be the same issue?

@csoceanu csoceanu Nov 3, 2026 •

Copy link
Copy Markdown
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

thanks, in fb05052 I clarified that an optional target code repository does not mean the run has no target entity. The source or invocation entity could still be a Jira issue, tenant context, or backlog item, and the run could report to that same entity or another explicit destination.

harness loading.
- Fullsend checks the field after harness resolution and capability
validation, but before environment validation, credential selection, script
execution, or sandbox creation.

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I think we might want something more broad here - perhaps a way to select the target entity kind with special flags for reusing the source entity or for the "none" case if we need it.

Copy link
Copy Markdown
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I agree that source/target entity selection is a broader design question. I kept it outside this ADR so AISDLC-146 remains focused on separating the execution directory from the optional target code repository. Reusing the source entity, choosing another entity kind, or having no target entity can be handled separately.

Co-authored-by: Codex <noreply@openai.com>
Signed-off-by: csoceanu <csoceanu@redhat.com>
@fullsend-ai-review

fullsend-ai-review Bot commented Nov 3, 2026 •

Copy link
Copy Markdown

🤖 Finished Review · ✅ Success · Started 5:10 PM UTC · Completed 5:19 PM UTC

Commit: fb05052 · View workflow run →

Runtime: pi · Model: sonnet → claude-sonnet-5-5 · Effort: high · Cost: $1.49

@fullsend-ai-review fullsend-ai-review Bot removed the ready-for-merge All reviewers approved — ready to merge label Nov 3, 2026
@csoceanu
csoceanu requested a review from ifireball November 3, 2026 17:12
@fullsend-ai-review
fullsend-ai-review Bot dismissed their stale review November 3, 2026 17:19

Superseded by updated review

@fullsend-ai-review fullsend-ai-review Bot added the ready-for-merge All reviewers approved — ready to merge label Nov 3, 2026

@rh-hemartin rh-hemartin left a comment

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Sounds good to me.

@ifireball ifireball left a comment

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

removing RC flag, I'm still not convinced we really need this, but no reason to block for now.

@rh-hemartin rh-hemartin left a comment

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Blocking this because some maintainers wanted to take a look at this today, just preventing a premature merge.

@csoceanu csoceanu closed this Nov 4, 2026

This branch was successfully deployed

1 active and 1 inactive deployments
site-preview — fb05052f Deployed Nov 3, 2026 by github-actions[bot]
dev — fb05052f Deployed Nov 3, 2026 by csoceanu via behaviour #18774
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

component/docs User-facing documentation fullsend-no-fix Skip bot-triggered fix agent runs ready-for-merge All reviewers approved — ready to merge risk/moderate PR risk: moderate

Projects

None yet

Development

Successfully merging this pull request may close these issues.

5 participants