Thanks to visit codestin.com
Credit goes to github.com

Skip to content

[Bug]: Amp dispatch sends -p/--model/--output-format — none of these exist in the Amp CLI #4580

Description

@hmohammadi

Bug Description

AmpIntegration never overrides build_exec_args(), so it inherits the generic implementation from MarkdownIntegration (src/specify_cli/integrations/base.py:1062), which emits:

<exe> -p <prompt> --model <model> --output-format json

None of those three flags exist in the Amp CLI. Every workflow command:/prompt: step dispatching to Amp fails immediately at argument parsing, even with amp installed and authenticated.

This is the amp row of the audit in #2416, which lists it first among the integrations still inheriting the unverified -p / --model / --output-format json pattern.

Steps to Reproduce

  1. Generate the argv Spec Kit dispatches with, from a source checkout at main (d848fb4):
$ uv run python -c "from specify_cli.integrations.amp import AmpIntegration; print(AmpIntegration().build_exec_args('hello', model='gpt-5', output_json=True))"
['amp', '-p', 'hello', '--model', 'gpt-5', '--output-format', 'json']
  1. Run exactly that against the real Amp CLI:
$ amp -p "hello" --model gpt-5 --output-format json
Error: error: unknown option '-p'
$ echo $?
1
  1. Confirm the flags Amp actually accepts:
$ amp --help | grep -E '^\s+-x|^\s+--stream-json$'
  -x, --execute [message]
  --stream-json

Expected Behavior

Spec Kit dispatches to Amp using Amp's documented execute-mode interface, so the command runs:

amp --execute <prompt> [--stream-json]

Actual Behavior

The dispatch fails at argument parsing with error: unknown option '-p' (exit 1). The agent is never invoked.

Checked against amp --help (v0.0.1789372854-ge954c8):

Flag Spec Kit sends Present in Amp CLI? Amp's actual flag
-p <prompt> No -x, --execute [message]
--model <model> No none — -m, --mode sets an agent mode (low/medium/high/ultra), not a model name
--output-format json No --stream-json (used with --execute)

Specify CLI Version

1.0.7.dev0 (source checkout, git d848fb4)

AI Agent

Amp

Operating System

Windows 11

Python Version

3.14.7

Error Logs

$ amp -p "hello" --model gpt-5 --output-format json
Error: error: unknown option '-p'

For contrast, the proposed argv parses and proceeds (it only stops at the login prompt, because this machine has no API key):

$ amp --execute "hello" --stream-json
No API key found. Starting login flow...
To log in, visit:
https://auth.ampcode.com/device?user_code=...

Additional Context

Amp CLI version tested: 0.0.1789372854-ge954c8, installed via npx @sourcegraph/amp.

Documentation: execute mode documents -x, --execute; streaming JSON states "Use --stream-json with --execute to output streaming JSON instead of plain text."

On the model parameter: Amp exposes no model-selection flag. -m, --mode takes an agent mode (low, medium, high, ultra, or a plugin mode) which "controls the model, system prompt, and tool selection" — it is not a model identifier, so passing Spec Kit's model value through to -m would be incorrect. A fix should drop the model flag rather than remap it.

Same fix shape as the existing one-off overrides for opencode (#2409) and goose (#3781). Happy to open a PR.

AI Assistance Disclosure: AI assistance was used for the source inspection, reproduction, and drafting of this report. All CLI output above was executed and captured on the reporting machine.

Activity

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    triage-nice-to-haveVerdict: evidence-backed fix or greenlit feature — land after review

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions