fix+test(#755): -p consumes exactly one token; flags after prompt text now parse normally

ROADMAP.md

@@ -7675,3 +7675,5 @@ Original filing (2026-04-18): the session emitted `SessionStart hook (completed)
 753. **`claw --output-format json -p` (no prompt arg) returned `error_kind:"unknown"` and `hint: null`** — parity gap with #750/#751 which fixed the explicit `prompt` verb. Identified by gaebal-gajae on `ddc71b56`. Fix: same `missing_prompt:` prefix + newline usage hint as #750. Regression guard: `short_p_flag_no_arg_json_error_kind_753` asserting nonzero exit, `error_kind:"missing_prompt"`, non-empty hint mentioning `claw -p` or `claw prompt`. Source: gaebal-gajae dogfood on `ddc71b56`, 2026-05-26.
 
 754. **`missing_credentials` JSON envelope always had `hint: null` even when a contextual hint was available** — dogfooded 2026-05-26 on `e9327135`. `ApiError::Display` for `MissingCredentials` appended the hint via ` — hint: {hint}` (inline, no `\n`), so `split_error_hint()` could not extract it and left the JSON `hint` field null. Fix: change delimiter from ` — hint: ` to `\n` in `api/src/error.rs` Display impl; update two tests in `api/src/error.rs` and `api/src/providers/mod.rs` to assert newline separator. Source: Jobdori dogfood on `e9327135`, 2026-05-26.
+
+755. **`claw -p hello --model sonnet` swallowed `--model sonnet` into the prompt string** — gaebal-gajae pinpoint on `e9327135` (#117 revival). `-p` used `args[index+1..].join(" ")`, consuming all remaining tokens as prompt. Fix: capture exactly one token via `args.get(index+1)`, reject flag-like tokens (`starts_with('-')`) as `missing_prompt`, support `--` sentinel for literal flag-text, then `continue` the flag loop so `--model`/`--output-format`/etc. parse normally. Dispatch via `short_p_prompt` after full flag scan. Regression guard: `short_p_flag_swallows_no_flags_755` asserts `--output-format json` is parsed (not swallowed) and `--model` as prompt-arg is rejected. Source: gaebal-gajae dogfood on `e9327135`, 2026-05-26.

rust/crates/rusty-claude-cli/src/main.rs

@@ -742,6 +742,9 @@ fn parse_args(args: &[String]) -> Result<CliAction, String> {
     let mut base_commit: Option<String> = None;
     let mut reasoning_effort: Option<String> = None;
     let mut allow_broad_cwd = false;
+    // #755: -p prompt text captured as single token; remaining args continue
+    // flag parsing. None until `-p <text>` is seen.
+    let mut short_p_prompt: Option<String> = None;
     let mut rest: Vec<String> = Vec::new();
     let mut index = 0;
 
@@ -858,24 +861,40 @@ fn parse_args(args: &[String]) -> Result<CliAction, String> {
                 index += 1;
             }
             "-p" => {
-                // Claw Code compat: -p "prompt" = one-shot prompt
+                // Claw Code compat: -p "prompt" = one-shot prompt.
-                let prompt = args[index + 1..].join(" ");
+                // #755: consume exactly one token so subsequent flags like
-                if prompt.trim().is_empty() {
+                // --model/--output-format are parsed normally instead of
-                    // #753: same missing_prompt shape as claw prompt (no arg) fix in #750
+                // being swallowed into the prompt string (#117).
-                    return Err("missing_prompt: -p requires a prompt string.\nUsage: claw -p <text>  or  claw prompt <text>".to_string());
+                let next = args.get(index + 1).map(|s| s.as_str());
+                match next {
+                    None | Some("") => {
+                        return Err("missing_prompt: -p requires a prompt string.\nUsage: claw -p <text>  or  claw prompt <text>".to_string());
+                    }
+                    Some(tok) if tok.starts_with('-') && tok != "--" => {
+                        // Looks like a flag, not a prompt. Reject so the user
+                        // knows to quote the literal text or use `--`.
+                        return Err(format!(
+                            "missing_prompt: -p requires a prompt string before flags; got `{tok}`.\nUsage: claw -p <text> --model sonnet  or  claw -p -- {tok} (literal)"
+                        ));
+                    }
+                    Some(tok) => {
+                        // `--` sentinel: skip it and take the token after as literal
+                        let (prompt_text, skip) = if tok == "--" {
+                            match args.get(index + 2) {
+                                Some(t) => (t.as_str(), 3usize),
+                                None => return Err("missing_prompt: -p -- requires a prompt string after `--`.\nUsage: claw -p -- <text>".to_string()),
+                            }
+                        } else {
+                            (tok, 2usize)
+                        };
+                        if prompt_text.trim().is_empty() {
+                            return Err("missing_prompt: -p requires a non-empty prompt string.\nUsage: claw -p <text>  or  claw prompt <text>".to_string());
+                        }
+                        short_p_prompt = Some(prompt_text.to_string());
+                        index += skip;
+                        continue;
+                    }
                 }
-                return Ok(CliAction::Prompt {
-                    prompt,
-                    model: resolve_model_alias_with_config(&model),
-                    output_format,
-                    allowed_tools: normalize_allowed_tools(&allowed_tool_values)?,
-                    permission_mode: permission_mode_override
-                        .unwrap_or_else(default_permission_mode),
-                    compact,
-                    base_commit: base_commit.clone(),
-                    reasoning_effort: reasoning_effort.clone(),
-                    allow_broad_cwd,
-                });
             }
             "--print" => {
                 // Claw Code compat: --print makes output non-interactive
@@ -965,6 +984,21 @@ fn parse_args(args: &[String]) -> Result<CliAction, String> {
 
     let allowed_tools = normalize_allowed_tools(&allowed_tool_values)?;
 
+    // #755: -p consumed exactly one token; dispatch now that all flags are parsed
+    if let Some(prompt) = short_p_prompt {
+        return Ok(CliAction::Prompt {
+            prompt,
+            model: resolve_model_alias_with_config(&model),
+            output_format,
+            allowed_tools,
+            permission_mode: permission_mode_override.unwrap_or_else(default_permission_mode),
+            compact,
+            base_commit,
+            reasoning_effort,
+            allow_broad_cwd,
+        });
+    }
+
     if rest.is_empty() {
         let permission_mode = permission_mode_override.unwrap_or_else(default_permission_mode);
         // When stdin is not a terminal (pipe/redirect) and no prompt is given on the

rust/crates/rusty-claude-cli/tests/output_format_contract.rs

@@ -1504,6 +1504,69 @@ fn prompt_no_arg_json_error_kind_750() {
     );
 }
 
+#[test]
+fn short_p_flag_swallows_no_flags_755() {
+    // #755: `claw -p hello --output-format json` must parse --output-format json
+    // as a flag rather than swallowing it as part of the prompt. Before #755,
+    // args[index+1..].join(" ") consumed all remaining tokens into the prompt.
+    // After #755, -p consumes exactly one token and remaining flags are parsed.
+    // We verify by checking that the envelope IS JSON (meaning --output-format json
+    // was interpreted as a flag, not literal prompt text).
+    use std::process::Command;
+    let root = unique_temp_dir("short-p-flags");
+    fs::create_dir_all(&root).expect("temp dir");
+    let bin = env!("CARGO_BIN_EXE_claw");
+
+    // -p hello --output-format json: with no credentials, should fail with
+    // missing_credentials (not missing_prompt), proving --output-format json was parsed.
+    let output = Command::new(bin)
+        .current_dir(&root)
+        .args(["-p", "hello", "--output-format", "json"])
+        .env_remove("ANTHROPIC_API_KEY")
+        .env_remove("ANTHROPIC_AUTH_TOKEN")
+        .output()
+        .expect("claw -p should run");
+    assert!(
+        !output.status.success(),
+        "claw -p hello --output-format json must exit non-zero (no credentials)"
+    );
+    let raw = String::from_utf8_lossy(&output.stderr)
+        .lines()
+        .filter(|l| l.starts_with('{'))
+        .collect::<Vec<_>>()
+        .join("");
+    // Must be valid JSON (i.e. --output-format json was parsed, not swallowed)
+    let parsed: serde_json::Value = serde_json::from_str(&raw).unwrap_or_else(|_| {
+        panic!("--output-format json must be parsed as a flag, not prompt text; stderr: {raw}")
+    });
+    assert_eq!(
+        parsed["error_kind"], "missing_credentials",
+        "flags after -p prompt text must be parsed normally (#755); got: {parsed}"
+    );
+
+    // Also verify -p --model bogus is rejected as missing_prompt (flag-as-prompt guard)
+    let output2 = Command::new(bin)
+        .current_dir(&root)
+        .args(["--output-format", "json", "-p", "--model", "sonnet"])
+        .output()
+        .expect("claw -p flag-as-prompt should run");
+    let raw2 = String::from_utf8_lossy(&output2.stderr)
+        .lines()
+        .filter(|l| l.starts_with('{'))
+        .collect::<Vec<_>>()
+        .join("");
+    let parsed2: serde_json::Value = serde_json::from_str(&raw2)
+        .unwrap_or_else(|_| panic!("claw -p --model must emit JSON; got: {raw2}"));
+    assert_eq!(
+        parsed2["error_kind"], "missing_prompt",
+        "flag-like token after -p must be rejected as missing_prompt (#755): {parsed2}"
+    );
+    assert!(
+        parsed2["hint"].as_str().map_or(false, |h| !h.is_empty()),
+        "missing_prompt hint must be non-empty (#755)"
+    );
+}
+
 #[test]
 fn short_p_flag_no_arg_json_error_kind_753() {
     // #753: `claw --output-format json -p` (no prompt) must emit error_kind:"missing_prompt"