`agents plan` hides results and gives very incomplete results. #10878

New issue

Closed

opened 2026-04-28 01:15:05 +00:00 by brent.edwards · 1 comment

brent.edwards commented

2026-04-28 01:15:05 +00:00

Member

Summary

Even with fixes, though Cleveragents claims to run plans, it hides its output.

Metadata

Commit message: fix(plan): output plan results
Branch: `bugfix/output-plan-results

Instructions

From a new docker exec for a CleverAgents docker container:

mkdir -p ~/temp/20260427-01
cd ~/temp/20260427-01
uv venv
source .venv/bin/activate
uv pip install /app
agents init --yes

Create the project

agents project create -d "The CleverAgents Project" local/cleveragents-project
agents resource add git-checkout local/cleveragents-resource --path /app
agents project link-resource local/cleveragents-project local/cleveragents-resource

Notice that the plan is similar to the existing one in /app/examples/actions/read-only.yaml, but:

It specifies the strategy-actor and the execution_actor (so that we don't need to create them for this simple test)
It changes read_only: true because of Issue #4194 .
The definition_of_done has been updated to request a complete, English-language report.

cat << EOF > review.yaml
name: local/architecture-review
description: "Review project architecture and produce a report"
long_description: |
  Analyzes the project structure, dependency graph, and code patterns
  to produce an architecture review report. Does not modify any files.

strategy_actor: anthropic/claude-sonnet-4-6
execution_actor: anthropic/claude-sonnet-4-6

definition_of_done: |
  Create a complete English-language architecture review report covering:
  - Module dependency graph
  - Layer boundary compliance
  - Circular dependency detection
  - Code duplication analysis
  Your output must contain all sections that you list in the table of contents.

reusable: true
read_only: false
state: available
EOF
agents action create --config review.yaml

Use the plan!

agents plan use local/architecture-review local/cleveragents-project

On my machine, this gives the following result:

╭──────────────────────────── Plan Created ─────────────────────────────╮
│ ID: 01KQ8QB6RZ2D8NW9NQ20DF9KKP                                        │
│ Name: local/architecture-review-01kq8qb6                              │
│ Action: local/architecture-review                                     │
│ Phase: strategize                                                     │
│ Processing State: queued                                              │
│ Projects: local/cleveragents-project                                  │
│ Description:                                                          │
│   Analyzes the project structure, dependency graph, and code patterns │
│ to produce an architecture review report. Does not modify any files.  │
│ Strategy Actor: anthropic/claude-sonnet-4-6                           │
│ Execution Actor: anthropic/claude-sonnet-4-6                          │
│ Definition of Done:                                                   │
│   Architecture review report generated covering:                      │
│ - Module dependency graph                                             │
│ - Layer boundary compliance                                           │
│ - Circular dependency detection                                       │
│ - Code duplication analysis                                           │
│ Automation Profile: manual (source: global)                           │
│ Terminal: no                                                          │
│ Created: 2026-04-28 00:20:38.815617                                   │
│ Updated: 2026-04-28 00:20:38.815618                                   │
╰───────────────────────────────────────────────────────────────────────╯

agents plan execute 01KQ8QB6RZ2D8NW9NQ20DF9KKP

On my machine, this takes some time to complete. Here is the result:

╭──────────────────────────── Plan Executed ────────────────────────────╮
│ ID: 01KQ8QB6RZ2D8NW9NQ20DF9KKP                                        │
│ Name: local/architecture-review-01kq8qb6                              │
│ Action: local/architecture-review                                     │
│ Phase: execute                                                        │
│ Processing State: complete                                            │
│ Projects: local/cleveragents-project                                  │
│ Description:                                                          │
│   Analyzes the project structure, dependency graph, and code patterns │
│ to produce an architecture review report. Does not modify any files.  │
│ Strategy Actor: anthropic/claude-sonnet-4-6                           │
│ Execution Actor: anthropic/claude-sonnet-4-6                          │
│ Definition of Done:                                                   │
│   Architecture review report generated covering:                      │
│ - Module dependency graph                                             │
│ - Layer boundary compliance                                           │
│ - Circular dependency detection                                       │
│ - Code duplication analysis                                           │
│ Automation Profile: manual (source: global)                           │
│ Last Checkpoint: 01KQ8QXCC5HAZCFKFHS7AEQXRZ                           │
│ Terminal: no                                                          │
│ Created: 2026-04-28 00:20:38.815617                                   │
│ Updated: 2026-04-28 00:30:34.384166                                   │
│ Strategize Started: 2026-04-28 00:22:19.149875                        │
│ Strategize Completed: 2026-04-28 00:23:00.693467                      │
│ Execute Started: 2026-04-28 00:23:00.755584                           │
│ Execute Completed: 2026-04-28 00:30:34.384128                         │
╰───────────────────────────────────────────────────────────────────────╯

Plan execution completed (execute/complete). Run 'agents plan apply <id>' when ready.

At this point, you can visit the directory /home/devuser/temp/20260427-01/.cleveragents/sandbox/architecture-review, and
there will exist a file architecture_review_report.md. So there IS a review. (I have enclosed the review.)

However, the review is extremely incomplete; the table of contents reports 9 sections, but it cuts off at the start of the third section. It is also quite wrong -- for example, it reports

**Critical:** A 4-module circular dependency chain involving `agent-core → orchestrator → tool-registry → agent-core` degrades testability and deployment flexibility.

But you can verify that there is no agent-core by typing

grep -R "agent.core" .

In addition, when I read through other files, it seems clear that the analysis is done on Typescript files instead of Python files. It seems very likely that the analysis had nothing to do with the actual source in /app.

Acceptance Criteria

The output must be easier for the average user to find.
The output must be based on the actual source.
The output must be complete.

# Summary Even with fixes, though Cleveragents claims to run plans, it hides its output. # Metadata - Commit message: `fix(plan): output plan results` - Branch: `bugfix/output-plan-results # Instructions From a new `docker exec` for a CleverAgents docker container: ``` mkdir -p ~/temp/20260427-01 cd ~/temp/20260427-01 uv venv source .venv/bin/activate uv pip install /app agents init --yes ``` Create the project ``` agents project create -d "The CleverAgents Project" local/cleveragents-project agents resource add git-checkout local/cleveragents-resource --path /app agents project link-resource local/cleveragents-project local/cleveragents-resource ``` Notice that the plan is similar to the existing one in [`/app/examples/actions/read-only.yaml`](https://git.cleverthis.com/cleveragents/cleveragents-core/src/branch/master/examples/actions/read-only.yaml), but: 1. It specifies the `strategy-actor` and the `execution_actor` (so that we don't need to create them for this simple test) 2. It changes `read_only: true` because of [Issue #4194](https://git.cleverthis.com/cleveragents/cleveragents-core/issues/4194) . 3. The `definition_of_done` has been updated to request a complete, English-language report. ``` cat << EOF > review.yaml name: local/architecture-review description: "Review project architecture and produce a report" long_description: | Analyzes the project structure, dependency graph, and code patterns to produce an architecture review report. Does not modify any files. strategy_actor: anthropic/claude-sonnet-4-6 execution_actor: anthropic/claude-sonnet-4-6 definition_of_done: | Create a complete English-language architecture review report covering: - Module dependency graph - Layer boundary compliance - Circular dependency detection - Code duplication analysis Your output must contain all sections that you list in the table of contents. reusable: true read_only: false state: available EOF agents action create --config review.yaml ``` Use the plan! ``` agents plan use local/architecture-review local/cleveragents-project ``` On my machine, this gives the following result: ``` ╭──────────────────────────── Plan Created ─────────────────────────────╮ │ ID: 01KQ8QB6RZ2D8NW9NQ20DF9KKP │ │ Name: local/architecture-review-01kq8qb6 │ │ Action: local/architecture-review │ │ Phase: strategize │ │ Processing State: queued │ │ Projects: local/cleveragents-project │ │ Description: │ │ Analyzes the project structure, dependency graph, and code patterns │ │ to produce an architecture review report. Does not modify any files. │ │ Strategy Actor: anthropic/claude-sonnet-4-6 │ │ Execution Actor: anthropic/claude-sonnet-4-6 │ │ Definition of Done: │ │ Architecture review report generated covering: │ │ - Module dependency graph │ │ - Layer boundary compliance │ │ - Circular dependency detection │ │ - Code duplication analysis │ │ Automation Profile: manual (source: global) │ │ Terminal: no │ │ Created: 2026-04-28 00:20:38.815617 │ │ Updated: 2026-04-28 00:20:38.815618 │ ╰───────────────────────────────────────────────────────────────────────╯ ``` ``` agents plan execute 01KQ8QB6RZ2D8NW9NQ20DF9KKP ``` On my machine, this takes some time to complete. Here is the result: ``` ╭──────────────────────────── Plan Executed ────────────────────────────╮ │ ID: 01KQ8QB6RZ2D8NW9NQ20DF9KKP │ │ Name: local/architecture-review-01kq8qb6 │ │ Action: local/architecture-review │ │ Phase: execute │ │ Processing State: complete │ │ Projects: local/cleveragents-project │ │ Description: │ │ Analyzes the project structure, dependency graph, and code patterns │ │ to produce an architecture review report. Does not modify any files. │ │ Strategy Actor: anthropic/claude-sonnet-4-6 │ │ Execution Actor: anthropic/claude-sonnet-4-6 │ │ Definition of Done: │ │ Architecture review report generated covering: │ │ - Module dependency graph │ │ - Layer boundary compliance │ │ - Circular dependency detection │ │ - Code duplication analysis │ │ Automation Profile: manual (source: global) │ │ Last Checkpoint: 01KQ8QXCC5HAZCFKFHS7AEQXRZ │ │ Terminal: no │ │ Created: 2026-04-28 00:20:38.815617 │ │ Updated: 2026-04-28 00:30:34.384166 │ │ Strategize Started: 2026-04-28 00:22:19.149875 │ │ Strategize Completed: 2026-04-28 00:23:00.693467 │ │ Execute Started: 2026-04-28 00:23:00.755584 │ │ Execute Completed: 2026-04-28 00:30:34.384128 │ ╰───────────────────────────────────────────────────────────────────────╯ Plan execution completed (execute/complete). Run 'agents plan apply <id>' when ready. ``` At this point, you can visit the directory `/home/devuser/temp/20260427-01/.cleveragents/sandbox/architecture-review`, and there will exist a file `architecture_review_report.md`. So there IS a review. (I have enclosed the review.) However, the review is extremely incomplete; the table of contents reports 9 sections, but it cuts off at the start of the third section. It is also quite wrong -- for example, it reports ``` **Critical:** A 4-module circular dependency chain involving `agent-core → orchestrator → tool-registry → agent-core` degrades testability and deployment flexibility. ``` But you can verify that there is no `agent-core` by typing ``` grep -R "agent.core" . ``` In addition, when I read through other files, it seems clear that the analysis is done on Typescript files instead of Python files. It seems very likely that the analysis had nothing to do with the actual source in `/app`. # Acceptance Criteria - The output must be easier for the average user to find. - The output must be based on the actual source. - The output must be complete.

Rows
Columns

agents plan hides results and gives very incomplete results. #10878

Summary

Metadata

Instructions

Acceptance Criteria

`agents plan` hides results and gives very incomplete results. #10878